RAD51C c.571+15G>T, p.?

Reasoning

PM2 (supporting) applies when the variant is absent from, or present below the expected carrier frequency in, a large general-population cohort; per the ClinGen SVI recommendation PM2 is used at supporting strength and the operative rarity threshold is <=0.0001 (0.01%) (PMID:25741868). RAD51C c.571+15G>T is present at a total AF of 1.61173e-05 (0.0016%; 26/1,613,178 alleles) in gnomAD v4.1 with a joint grpmax FAF of 1.377e-05, and at 1.59129e-05 (4/251,368 alleles) with grpmax FAF 7.01e-06 in gnomAD v2.1 exomes. Both matching non-cancer subsets are equally rare (v2.1 non-cancer exomes 1.68898e-05 with grpmax FAF 7.76e-06; v3.1 non-cancer genomes 2.02807e-05 with grpmax FAF 1.972e-05). Every dataset total and every ancestry-adjusted grpmax FAF is at or below 1.4e-05, comfortably under 0.0001. The only values exceeding 0.0001 are single-allele bins in the small 'Remaining individuals' ancestry group (1.63132e-04 = 1/6,130 in v2.1 and 1.78253e-04 = 1/5,610 in v2.1 non-cancer exomes), which are sampling noise in ~5,000-6,000-allele bins rather than a meaningful signal of an ancestry-specific common allele, and the ancestry-adjusted grpmax FAF for those same datasets remains ~1e-05. The variant is not absent from gnomAD, so PM2 is applied on rarity (well below the threshold) rather than on absence, and only at supporting strength. PM2 is met.

Rule

PM2: supporting pathogenic when allele frequency <= 0.0001 (SVI downgrade to supporting, PMID:25741868)

Evidence considered

• gnomAD v4.1 all-comers (default source): total AF 1.61173e-05 (0.0016%), 26/1,613,178 alleles, 1 homozygote; joint grpmax FAF 1.377e-05; highest ancestry frequency European (non-Finnish) 2.03507e-05 (24/1,179,318).
• gnomAD v2.1 exomes (default source): AF 1.59129e-05 (0.0016%), 4/251,368 alleles, 0 homozygotes; grpmax FAF 7.01e-06.
• gnomAD v2.1 non-cancer exomes: AF 1.68898e-05, 4/236,830 alleles, 0 homozygotes; grpmax FAF 7.76e-06; largest ancestry bin 'Remaining individuals' 1.78253e-04 (1/5,610 alleles).
• gnomAD v3.1 non-cancer genomes: AF 2.02807e-05, 3/147,924 alleles, 0 homozygotes; grpmax FAF 1.972e-05; highest ancestry frequency European (non-Finnish) 4.62863e-05 (3/64,814 alleles).
• PM2 threshold applied: allele frequency <=0.0001 (0.01%), SVI recommendation to downgrade PM2 to supporting strength (PMID:25741868). All dataset-level and ancestry-adjusted grpmax values (<=1.4e-05) are below this threshold.
• Threshold-artifact check: the only frequencies above 0.0001 are single-allele observations in ~5,000-6,000-allele 'Remaining individuals' bins (1.63e-04 and 1.78e-04); a single allele in a bin this size is not a reliable ancestry-specific frequency estimate, and the ancestry-adjusted grpmax FAF for both datasets remains ~1e-05, i.e. below the threshold.
• The variant is present rather than absent in gnomAD, so PM2 rests on rarity below the expected carrier frequency for this rare disorder, which the ACMG/AMP 2015 population-data section explicitly permits for variants below the expected carrier frequency (PMID:25741868).
• gnomAD-Canada (HostSeq) errored and returned no data; it contributed nothing to this assessment.

Still missing

• The RAD51C HBOP VCEP specification was not machine-readable in this case, so it could not be confirmed whether that expert panel retains PM2 for RAD51C or imposes a different rarity threshold; the generic SVI supporting-strength threshold (<=0.0001) was applied.
• The small 'Remaining individuals' ancestry bins return frequencies above 0.0001 from a single allele each; if the governing specification requires a popmax-AF rather than grpmax-FAF metric, a reviewer should confirm the threshold comparison for that bin.

Sources

Reasoning

Same scope determination as PP3: NM_058216.3:c.571+15G>T is an intronic variant (intron 3, 15 nucleotides from the exon 3 boundary; p.? on NM_058216.3), so BP4 is taken from the SpliceAI path only. SpliceAI Lookup (hg37, basic, distance 500, mask 0) returned DS_AG 0.003, DS_AL 0.000, DS_DG 0.003, DS_DL 0.006, max delta 0.006; the largest value sits well below the published BP4 supporting cutoff of <= 0.1 (Jaganathan et al. 2019, Cell, PMID:30661751), and no donor-acceptor gain or loss is predicted at any position in the 500 bp window. BP4 is therefore met at supporting strength. One internal inconsistency was noted and is not verdict-changing: the bundle's narrative evidence sentence quotes 'max delta score = 0.01', while the structured SpliceAI per-position scores give a maximum of 0.006 (0.01 corresponds to the Pangolin SG value in the same record). Both candidate values are <= 0.1, so the conclusion is the same either way; the structured per-position SpliceAI scores were used as the primary metric. The HBOP/RAD51C VCEP specification v1.0 contains no gene-specific BP4 rule or lookup table, so the published generic cutoff governs. REVEL was not used here: this is not a missense variant, its REVEL lookup returned no score, and a splice-path benign prediction is the correct basis for BP4 for an intronic change.

Rule

SpliceAI max delta <= 0.1 -> BP4 supporting (Jaganathan et al. 2019, Cell; PMID:30661751). Generic rule (no gene-specific VCEP rule available).

Evidence considered

• SpliceAI Lookup for NM_058216.3:c.571+15G>T (GRCh37, basic, distance 500, mask 0): DS_AG 0.003, DS_AL 0.000, DS_DG 0.003, DS_DL 0.006; max delta score 0.006; no predicted gain/loss of donor or acceptor site (DP_AG -174, DP_AL -82, DP_DG -15, DP_DL -129).
• Threshold applied for BP4 on the splice path: SpliceAI max delta <= 0.1 for BP4 supporting (Jaganathan et al. 2019, Cell; PMID:30661751). Max delta 0.006 satisfies this.
• Score discrepancy documented: the narrative evidence sentence in the case record states 'max delta score = 0.01', whereas the structured SpliceAI scores in the same record give a maximum of 0.006 (0.01 matches the Pangolin SG field); both values are <= 0.1 and the BP4 conclusion is unchanged.
• The HBOP VCEP RAD51C specification v1.0 (CSPEC doc 1535377178) was reviewed first and provides no gene-specific BP4/PP3 scoring rules, so the published generic SpliceAI cutoff was applied.
• REVEL was not consulted as support for BP4: the variant is intronic (not missense) and no REVEL score was retrieved for it.

Sources

Reasoning

PM1 requires the variant to lie in a mutational hotspot and/or a critical, well-established functional domain without benign variation. The authoritative VCEP domain resource for this gene is absent: vcep_materials.json 'domain_tables' is an empty object ({}), so there is no RAD51C-approved critical-domain table against which the variant's residue could be checked, and the governing ClinGen HBOP/RAD51C VCEP specification v1.0 (CSPEC doc 1535377178, transcript NM_058216.3) carries an empty criteria payload with no PM1 rule or domain definition. The variant is nevertheless intronic - NM_058216.3:c.571+15G>T lies 15 nucleotides into intron 3 (VariantValidator exon context '3i') with predicted protein consequence NP_478123.1:p.? - so it encodes no amino-acid residue and cannot lie within any critical functional domain or mutational hotspot of the RAD51C protein; there is accordingly no residue to test against any domain entry, approved or otherwise. Independent hotspot sources likewise returned nothing for this variant: COSMIC returned no result, the case-level hotspot record is empty (found null), and the OncoKB gene page failed to load, so no hotspot assignment exists. SpliceAI max delta 0.006 (DS_AG 0.003, DS_AL 0.000, DS_DG 0.003, DS_DL 0.006; no donor or acceptor gain/loss in the 500 bp window) also shows that the change does not create or disrupt a splice-region element, so it cannot be counted as a splice-region hotspot. PM1 is therefore not met.

Rule

Generic ACMG/AMP PM1 (mutational hotspot and/or critical functional domain without benign variation; PMID:25741868) - not met: the variant is intronic with no encoded residue, outside any RAD51C domain or hotspot, and no gene-specific PM1 rule exists.

Evidence considered

• vcep_materials.json contains an empty domain_tables object ({}), i.e. no RAD51C critical-functional-domain list extracted from the governing VCEP domain table - there is no approved entry to check the variant's residue against.
• final_classification_framework.json shows the governing framework for this case is the CSPEC ruleset for RAD51C v1.0 (doc 1535377178), recorded as framework_mode 'unstructured_ruleset' with framework_complete false and an empty rule_payload, so no gene-specific PM1 rule exists.
• VariantValidator normalization places NM_058216.3:c.571+15G>T in intron 3 (variant_exonic_positions start/end exon '3i') with predicted protein consequence NP_478123.1:p.?, so the variant encodes no residue of the RAD51C protein and cannot occupy a mutational hotspot or critical functional domain.
• SpliceAI Lookup for NM_058216.3:c.571+15G>T (GRCh37, basic, distance 500, mask 0): DS_AG 0.003, DS_AL 0.000, DS_DG 0.003, DS_DL 0.006, max delta score 0.006, with predicted positions DP_AG -174, DP_AL -82, DP_DG -15, DP_DL -129 - no gain or loss of a donor/acceptor site, so no splice-region hotspot is created.
• COSMIC returned 'no result' for RAD51C c.571+15G>T and the case hotspot record is empty (hotspots found null), so no external hotspot assignment is available for this position.
• The ClinVar record for this variant (variation 492371) carries four clinical-laboratory submissions, all classified Likely benign with no expert-panel review; no submitter provides domain or hotspot reasoning, and no criterion was inferred from any submitter's classification.
• PMID:25741868 supplies the generic ACMG/AMP PM1 definition (located in a mutational hot spot and/or critical and well-established functional domain without benign variation) applied here; it contains no RAD51C data.

Still missing

• vcep_materials.json 'domain_tables' is empty ({}), so no VCEP-approved RAD51C critical-domain boundary list was available to check the variant against; the PM1 conclusion rests on the variant being intronic with no encoded residue, which makes domain membership impossible for any domain definition.
• The HBOP/RAD51C VCEP specification (CSPEC doc 1535377178) present in this case has an empty criteria payload, so no gene-specific PM1 rule, hotspot list or splice-region PM1 provision could be applied; generic ACMG/AMP PM1 governed.
• No hotspot data source yielded a result for this variant: COSMIC returned 'no result', the case-level hotspots object is empty (found null), and the OncoKB page failed to load (h1 selector timeout), so hotspot assessment rests on the intronic location and the SpliceAI result rather than on a hotspot database.

Sources

Reasoning

PM3 applies to recessive disorders and requires that this variant be detected in trans with a pathogenic variant - i.e. a proband carrying a confirmed pathogenic allele on the opposite homolog. RAD51C has a bona fide recessive association (biallelic loss causes Fanconi anemia complementation group O, MONDO:0019391), alongside its dominant association with familial ovarian/breast cancer, so the criterion is conceptually applicable to this gene rather than categorically inapplicable. However, PM3 is a positive-evidence criterion and no such observation exists anywhere in the evidence assembled for this case. NM_058216.3:c.571+15G>T (NC_000017.10:g.56774235G>T; NC_000017.11:g.58696874G>T) is a deep intronic substitution 15 nucleotides into intron 3 (variant_exonic_positions 3i; predicted protein consequence NP_478123.1:p.?), and the case carries no proband-level data at all: no affected individual, no second RAD51C variant, no zygosity and no phasing information, and no segregation/family study. Screening of the four ClinVar submissions for VCV000492371 (GeneDx, Color Diagnostics, Labcorp/Invitae, Department of Pathology and Laboratory Medicine Sinai Health System) found all four to be single-submitter Likely benign records with no expert-panel review and no submission criterion lead of the trans/cis type; the submission audit surfaced only one weak in-silico-derived lead (an in-silico, likely-benign-style statement) and no PM3-type evidence. The literature retrieved for this case provides nothing either: the two papers with full text available (PMID:25741868, the ACMG/AMP framework, and PMID:25394175, the ACMG/NSGC cancer-referral guideline) never name RAD51C or this variant, and neither the Fanconi Anemia GeneReviews chapter (PMID:20301575, flagged high priority for case observations) nor the Sherloc criteria paper (PMID:28492532) has a retrieved full text or pre-computed extraction in this case, so no affected individual with this variant in trans with a pathogenic RAD51C allele can be documented from any consulted source. Absent a documented in-trans pathogenic partner allele in an affected proband, PM3 is not met at any strength. No RAD51C-specific modification is available: the nominally governing ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer specification for RAD51C v1.0 (doc 1535377178) is present but exposes no criterion-level rule payload, so the generic ACMG/AMP 2015 PM3 definition was applied.

Rule

ACMG/AMP 2015 PM3 (PMID:25741868): for recessive disorders, detected in trans with a pathogenic variant. No RAD51C-specific VCEP rule text available (CSPEC v1.0 doc 1535377178 rule payload empty). Verdict: not met - the case contains no proband with this variant in trans with a pathogenic RAD51C allele, and no consulted source reports such a configuration.

Evidence considered

• Variant identity and consequence: NM_058216.3:c.571+15G>T (NC_000017.10:g.56774235G>T; NC_000017.11:g.58696874G>T), annotated by VariantValidator as intronic with variant_exonic_positions 3i on MANE Select transcript NM_058216.3 (NP_478123.1, p.?), i.e. 15 nucleotides into intron 3 - a variant type that cannot itself establish or exclude an allelic (trans/cis) relationship.
• Gene inheritance context relevant to PM3: the RAD51C framework record lists both autosomal dominant diseases (familial ovarian cancer MONDO:0016248; hereditary breast carcinoma MONDO:0016419) and an autosomal recessive disease (Fanconi anemia MONDO:0019391), so PM3's recessive arm is applicable to this gene in principle.
• Case evidence contains no proband, no affected individual, no second RAD51C variant, no zygosity call, no phase/long-read data, and no segregation study: a full-term search of the case evidence for proband, phase, cis/trans, compound-heterozygous, and segregation content returned nothing.
• ClinVar VCV000492371 (variation 492371) reports this exact variant as Likely benign from four single-submitter laboratory records (GeneDx SCV000729555; Color Diagnostics SCV000691247; Labcorp/Invitae SCV002397711; Sinai Health System SCV005918798), all germline, no expert-panel submissions; the submission audit identified zero criterion leads of the trans/cis type, and the only evidence signal extracted was an in-silico, likely-benign-style statement from GeneDx.
• ClinVar final classifications were used only as context and lead screening; they were not used to infer PM3 or BP2.
• Literature scope: four PMIDs were assembled for this case (PMID:25741868, PMID:25394175, PMID:20301575, PMID:28492532). Full texts were retrieved for PMID:25741868 and PMID:25394175 only, and neither names RAD51C nor the variant, so neither can carry a trans observation for this variant. PMID:20301575 and PMID:28492532 have no full text and no pre-computed extraction in this case.
• Criterion definitions used: ACMG/AMP 2015 (PMID:25741868) defines PM3 as detection of the variant in trans with a pathogenic variant for a recessive disorder. The retrieved copy of that paper is truncated at 40,000 characters and ends before Table 3, so the criterion is not read from a quoted table row; what the retrieved copy does state verbatim is that 'rule PM3 could be upgraded to strong if there were multiple observations of detection of the variant in trans (on opposite chromosomes) with other pathogenic variants', and that classification may use unpublished case data obtained through public resources such as ClinVar or locus-specific databases - i.e. the trans observation itself is the evidence PM3 requires. No such observation exists for this variant.
• Governing framework check: the ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer Expert Panel specification for RAD51C v1.0 (CSPEC doc 1535377178, source key 'cspec') is present in the case, but its extracted rule payload is empty and the criterion list is empty, so no RAD51C-specific PM3 definition or threshold exists to override the generic ACMG/AMP 2015 text.
• Population data were not used for PM3: PM3 has no frequency component and no calibration threshold applies to it.

Still missing

• No proband, zygosity, or phase-resolved genotype data exist for this case, so a trans configuration with a pathogenic RAD51C allele could not be tested; PM3 would become assessable only if a future report or phased family study places c.571+15G>T in trans with a pathogenic RAD51C variant in an individual affected with Fanconi anemia complementation group O.
• PMID:20301575 (Fanconi Anemia, GeneReviews) - the one high-priority case-observation source that could plausibly describe biallelic RAD51C/FA-O cases - was not retrieved as full text in this case (only a screenshot image exists), so it could not be screened for this exact variant.
• The ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer RAD51C specification (v1.0, doc 1535377178) contains no retrievable criterion-level PM3 rule text, so any gene-specific PM3 strength modification or restriction (for example a requirement for confirmed biallelic status in FA) could not be applied.

Sources

Reasoning

Scope check first: NM_058216.3:c.571+15G>T in RAD51C is an intronic variant (VariantValidator places it within intron 3, start/end exon '3i', 15 nucleotides 3' of the exon 3 boundary; predicted protein consequence NP_478123.1:p.? on the MANE Select transcript, which is also the CSPEC-preferred transcript NM_058216.3). It is not missense and not truncating, so the applicable computational path is SpliceAI; REVEL is not evaluated for this variant and no REVEL score exists for it (lookup returned no score). SpliceAI Lookup (hg37, basic, distance 500, mask 0) returned DS_AG = 0.003, DS_AL = 0.000, DS_DG = 0.003, DS_DL = 0.006, with max delta score 0.006; the predicted positions are DP_AG -174, DP_AL -82, DP_DG -15, DP_DL -129, i.e. no gain or loss of a donor/acceptor site anywhere in the window, and variant_consequence was reported as sequence_variant. Against the published SpliceAI cutoffs (supporting PP3 >= 0.2; moderate PP3 >= 0.5; Jaganathan et al. 2019, Cell, PMID:30661751), a max delta of 0.006 does not reach even the supporting threshold, so PP3 is not met. The governing specification was checked first: the ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer (HBOP) VCEP specification for RAD51C v1.0 (CSPEC doc 1535377178) is present but carries an empty criteria/rules payload and no gene-specific PP3/BP4 lookup table, so the published generic SpliceAI thresholds were applied. The GeneDx ClinVar submission (SCV000729555) lists PP3 among its submitted criteria, but its in-silico sentence is a benign-oriented statement and no predictor score or direction supporting a damaging effect is given; no criterion was inferred from any submitter's final classification, and this does not alter the direct SpliceAI result.

Rule

SpliceAI max delta >= 0.2 -> PP3 supporting; >= 0.5 -> PP3 moderate (Jaganathan et al. 2019, Cell; PMID:30661751). Generic rule (no gene-specific VCEP rule available).

Evidence considered

• Variant normalization: NM_058216.3:c.571+15G>T, RAD51C, intronic (variant_exonic_positions start_exon/end_exon = '3i'), protein consequence NP_478123.1:p.?, MANE Select equals the CSPEC-preferred transcript NM_058216.3.
• SpliceAI Lookup result for NM_058216.3:c.571+15G>T (GRCh37, basic, distance 500, mask 0): DS_AG 0.003, DS_AL 0.000, DS_DG 0.003, DS_DL 0.006; max delta score 0.006; DP_AG -174, DP_AL -82, DP_DG -15, DP_DL -129; variant_consequence 'sequence_variant'; Pangolin SG 0.01, SL -0.006.
• Threshold applied for PP3 on the splice path: SpliceAI max delta >= 0.2 for supporting and >= 0.5 for moderate (Jaganathan et al. 2019, Cell; PMID:30661751). Max delta 0.006 reaches neither.
• The HBOP VCEP RAD51C specification v1.0 (CSPEC doc 1535377178) was consulted first and contains an empty criteria/rules payload with no gene-specific computational-evidence (PP3/BP4) scoring, so the published generic SpliceAI cutoffs govern.
• A ClinVar submission on this variant (GeneDx, SCV000729555) lists PP3 among submitted criteria; the underlying in-silico text is a benign-oriented statement with no score, so it was not treated as PP3 support, and no criterion was inferred from any submitter's final classification.
• REVEL is not the applicable path for this intronic variant and no REVEL score was available (lookup found no entry); it was not used to support or refute PP3.

Still missing

• No gene-specific PP3 rule or in-silico lookup table is present in the HBOP/RAD51C VCEP specification payload, so generic published SpliceAI cutoffs were used.

Sources

Reasoning

PP5 is an assertion-based criterion: it requires a reputable source that has already asserted the variant to be pathogenic, and under the rule governing this group that means an exact-variant ClinGen expert-panel classification of Pathogenic or Likely pathogenic - ordinary laboratory submissions, aggregate ClinVar labels and non-expert assertions must never trigger it. The exact-variant ClinVar record VCV000492371 (NM_058216.3(RAD51C):c.571+15G>T; exact match by transcript HGVS) has four submissions (GeneDx, Color Health, Labcorp/Invitae, Sinai Health) and every one is an ordinary clinical-testing laboratory submission: the submission audit records is_expert_panel = false for all four and expert_panel_submissions = 0. The record's review status is 'criteria provided, multiple submitters, no conflicts' (2 stars), which is not 'reviewed by expert panel' (3 stars) and therefore carries no expert-panel assertion. In addition to there being no expert-panel assertion, the direction of the available assertions is benign (all four laboratories classify the variant Likely benign), so there is no Pathogenic or Likely pathogenic assertion of any provenance to draw on - the aggregate ClinVar label cannot be repurposed as PP5 support, and no criterion in this group was inferred from any submitter's final classification. The governing HBOP/RAD51C VCEP specification v1.0 (CSPEC doc 1535377178) is present but its criteria payload is empty, so it supplies no gene-specific PP5 rule that could relax or replace this requirement. PP5 therefore carries no assumption of direction or threshold and, with the expert-panel precondition verifiably absent, it is not met.

Rule

PP5 (ACMG/AMP; Richards et al. 2015, PMID:25741868): a reputable source asserts the variant is pathogenic - operationalised for this group as an exact-variant ClinGen expert-panel Pathogenic or Likely pathogenic classification only. Ordinary laboratory submissions, aggregate ClinVar labels and non-expert assertions never trigger PP5.

Evidence considered

• Exact-variant ClinVar record VCV000492371 for NM_058216.3(RAD51C):c.571+15G>T: four submissions (SCV000729555 GeneDx, 2017-08-14; SCV000691247 Color Diagnostics/Color Health, 2015-12-02; SCV002397711 Labcorp Genetics (formerly Invitae), 2025-12-10; SCV005918798 Department of Pathology and Laboratory Medicine, Sinai Health System, 2022-11-25), all Likely benign, all germline clinical testing, all 'criteria provided, single submitter'.
• ClinVar aggregate review status for the record is 'criteria provided, multiple submitters, no conflicts' (2 stars), not 'reviewed by expert panel' (3 stars); the case summary likewise records expert_panels = {} for the variant.
• The submission audit records expert_panel_submissions = 0, and is_expert_panel = false for all four submissions, so no submission qualifies as a ClinGen expert-panel assertion.
• Per the rule governing this group, an exact-variant expert-panel Pathogenic/Likely pathogenic classification is the only trigger for PP5; ordinary laboratory submissions, aggregate ClinVar labels and non-expert assertions must never trigger it. That precondition is verifiably not satisfied.
• The HBOP VCEP RAD51C specification v1.0 (CSPEC doc 1535377178) is the governing framework and is present in the case as a reference, but its criteria/rules payload is empty (framework_complete=false, rule_payload={}), so no gene-specific PP5 rule exists to override the general requirement.
• PP5 was not inferred from any other source's classification, and no numeric threshold or comparison direction is attached to PP5, so no inequality was constructed.

Still missing

• No ClinGen expert-panel (3-star) classification exists for RAD51C c.571+15G>T in ClinVar; if the HBOP VCEP later classifies this variant, PP5/BP6 would need to be re-evaluated against that assertion.

Sources

Reasoning

BA1 requires a stand-alone benign allele frequency of >=0.05 (5%) per ACMG/AMP 2015 (PMID:25741868). RAD51C c.571+15G>T (NC_000017.10:g.56774235G>T) is present at a total AF of 1.61173e-05 (26/1,613,178 alleles, 1 homozygote) in gnomAD v4.1 and 1.59129e-05 (4/251,368 alleles, 0 homozygotes) in gnomAD v2.1 exomes, with ancestry-adjusted grpmax FAF of 1.377e-05 (v4.1) and 7.01e-06 (v2.1). The non-cancer subsets are concordant and equally rare (v2.1 non-cancer exomes 4/236,830 = 1.68898e-05, 0 homozygotes; v3.1 non-cancer genomes 3/147,924 = 2.02807e-05, 0 homozygotes). The highest frequency in any ancestry bin is 1.78253e-04 (1/5,610 alleles, 'Remaining individuals', v2.1 non-cancer exomes), roughly 280-fold below the 5% BA1 threshold, and that bin rests on a single allele. No dataset, population, or ancestry bin approaches 5%, so BA1 is not met.

Rule

BA1: stand-alone benign when allele frequency >= 0.05 (PMID:25741868)

Evidence considered

• gnomAD v4.1 all-comers (default source): total AF 1.61173e-05, 26/1,613,178 alleles, 1 homozygote; highest ancestry frequency European (non-Finnish) 2.03507e-05 (24/1,179,318 alleles, 1 homozygote); grpmax FAF 1.377e-05.
• gnomAD v2.1 exomes (default source): AF 1.59129e-05, 4/251,368 alleles, 0 homozygotes; highest subpopulation 'Remaining individuals' 1.63132e-04 (1/6,130 alleles, 0 homozygotes); grpmax FAF 7.01e-06.
• gnomAD v2.1 non-cancer exomes (exome-only figures): AF 1.68898e-05, 4/236,830 alleles, 0 homozygotes; highest subpopulation 'Remaining individuals' 1.78253e-04 (1/5,610 alleles); grpmax FAF 7.76e-06.
• gnomAD v3.1 non-cancer genomes (v3 has no exome cohort): AF 2.02807e-05, 3/147,924 alleles, 0 homozygotes; highest ancestry European (non-Finnish) 4.62863e-05 (3/64,814 alleles); grpmax FAF 1.972e-05.
• gnomAD-Canada (HostSeq) query errored and returned no data, so it could not contribute to the frequency assessment.
• BA1 threshold applied: allele frequency >=0.05 (5%), stand-alone benign evidence, per ACMG/AMP 2015 standards (Richards et al., PMID:25741868). The highest frequency observed anywhere (1.78e-04, one allele in 5,610) is approximately 280-fold below this threshold.
• The RAD51C HBOP VCEP specification is available as a document but carries no machine-readable population-criteria rule in this case, so the generic ACMG/AMP 2015 stand-alone threshold was used rather than a VCEP-specific value.

Still missing

• The RAD51C HBOP VCEP v1.0 specification (cspec doc 1535377178) is present for this gene but its ruleset payload is empty/unstructured in this case, so no VCEP-specific BA1 frequency threshold or permitted population source could be applied; the generic ACMG/AMP 2015 stand-alone threshold (5%) was used.

Sources

Reasoning

BS1 (strong benign) requires an allele frequency greater than that expected for the disorder, applied here as the generic >=0.01 threshold (Whiffin et al. 2017, PMID:28518168). The variant is rare in every source examined: gnomAD v4.1 total AF 1.61173e-05 (26/1,613,178 alleles), gnomAD v2.1 exomes 1.59129e-05 (4/251,368), gnomAD v2.1 non-cancer exomes 1.68898e-05 (4/236,830), gnomAD v3.1 non-cancer genomes 2.02807e-05 (3/147,924), with ancestry-adjusted grpmax FAF values of 1.377e-05 (v4.1) and 7.01e-06 (v2.1). The largest ancestry-bin frequency is 1.78253e-04 (1/5,610, 'Remaining individuals'), still ~56-fold below 1% and derived from a single allele. No source approaches the 0.01 BS1 threshold, and the observed frequency is consistent with the low prevalence of RAD51C-associated hereditary breast/ovarian cancer rather than with a benign polymorphism, so BS1 is not met.

Rule

BS1: strong benign when allele frequency >= 0.01 (Whiffin et al. 2017, PMID:28518168)

Evidence considered

• gnomAD v4.1 all-comers: total AF 1.61173e-05 (26/1,613,178 alleles, 1 homozygote); grpmax FAF 1.377e-05; highest ancestry frequency 2.03507e-05 (European non-Finnish, 24/1,179,318).
• gnomAD v2.1 exomes: AF 1.59129e-05 (4/251,368 alleles, 0 homozygotes); grpmax FAF 7.01e-06; largest ancestry bin 1.63132e-04 (1/6,130, 'Remaining individuals').
• gnomAD v2.1 non-cancer exomes: AF 1.68898e-05 (4/236,830 alleles, 0 homozygotes); grpmax FAF 7.76e-06; largest ancestry bin 1.78253e-04 (1/5,610).
• gnomAD v3.1 non-cancer genomes: AF 2.02807e-05 (3/147,924 alleles, 0 homozygotes); grpmax FAF 1.972e-05; highest ancestry frequency 4.62863e-05 (European non-Finnish, 3/64,814).
• BS1 threshold applied: allele frequency >=0.01 (1%), generic SVI-style default with per-gene derivation preferred per Whiffin et al. 2017 (PMID:28518168). Every dataset total and every ancestry bin is below 1%; the largest (1.78e-04) is roughly 56-fold lower.
• The RAD51C HBOP VCEP specification (cspec doc 1535377178) is present but contains no extracted population rule, so neither a gene-specific BS1 threshold nor a mandated non-cancer/exome-only data source could be applied.

Still missing

• No extraction of the RAD51C HBOP VCEP specification was available to confirm whether that expert panel prescribes a gene-specific BS1 threshold or a restricted population source (e.g. non-cancer/exome-only cohorts); the generic >=0.01 default was applied.

Sources

Reasoning

ACMG/AMP 2015 defines BS2 as strong benign evidence only when the variant is observed in a well-documented healthy adult and the disorder under investigation is fully penetrant at an early age (PMID:25741868). For RAD51C the dominant phenotype is adult-onset familial ovarian/breast cancer with reduced penetrance, where gnomAD adults cannot be treated as documented unaffected controls for that phenotype, so the BS2 precondition for the dominant condition is not satisfied. For the recessive, fully penetrant, early-onset Fanconi anemia phenotype, gnomAD v4.1 reports a single homozygous individual (1 homozygote among 26 total alleles; the homozygote sits in the European non-Finnish, XX subset, 18/611,220), while gnomAD v2.1, v2.1 non-cancer exomes and v3.1 non-cancer genomes report zero homozygotes. A single unconfirmed population-database homozygote call, with no documentation of adult health status or phenotype, does not meet the 'well-documented healthy adult individual' standard that BS2 requires, and no case-control or unaffected-adult frequency data were identified in the literature reviewed. BS2 is therefore not met.

Rule

BS2: strong benign when observed in a well-documented healthy adult for a disorder fully penetrant at an early age (PMID:25741868)

Evidence considered

• gnomAD v4.1: 1 reported homozygote overall (exome homozygote count 1, genome 0), located in the European (non-Finnish) XX subset (18 alleles in 611,220, homozygote_count 1); total 26/1,613,178 alleles.
• gnomAD v2.1 exomes: 0 homozygotes (4/251,368 alleles). gnomAD v2.1 non-cancer exomes: 0 homozygotes (4/236,830). gnomAD v3.1 non-cancer genomes: 0 homozygotes (3/147,924).
• BS2 definition applied: strong benign evidence requires the variant to be observed in a well-documented healthy adult individual AND the disease to be fully penetrant at an early age (Richards et al., ACMG/AMP 2015, PMID:25741868).
• Disease context: heterozygous germline RAD51C variants are associated with adult-onset familial ovarian cancer and hereditary breast carcinoma with reduced penetrance (autosomal dominant, as listed in the gene's ClinGen disease annotations), while biallelic RAD51C variants cause Fanconi anemia complementation group O (autosomal recessive, early onset).
• gnomAD participants are not a well-phenotyped control cohort for cancer predisposition; adult gnomAD carriers are not documented as unaffected, and reduced-penetrance adult-onset disease makes carrier status uninformative for BS2.
• No PMID in the case literature set (PMID:25741868, PMID:25394175) reports healthy-adult or case-control observations of this specific variant; paper_extracts entries for both confirm the variant is not mentioned.

Still missing

• Whether a single gnomAD v4.1 homozygous observation may count as BS2 evidence in this gene, and at what strength, depends on the RAD51C HBOP VCEP BS2 guidance, which is not machine-readable in this case.
• No case-control or well-phenotyped unaffected-adult data for c.571+15G>T were identified; the assessment rests on aggregate gnomAD homozygote counts only.
• The homozygote call in gnomAD could not be orthogonally confirmed (no read-level or pedigree data available), so its technical validity is unverified.

Sources

Reasoning

BP2 is satisfied by one of two phased observations: the variant is seen in trans with a pathogenic variant for a fully penetrant dominant gene/disorder, or it is seen in cis with a pathogenic variant in any inheritance pattern. RAD51C supports both inheritance modes (autosomal dominant familial ovarian/breast cancer; autosomal recessive Fanconi anemia complementation group O), so both arms of BP2 are conceptually open for this gene, but neither is fulfilled for NM_058216.3:c.571+15G>T (NC_000017.10:g.56774235G>T). There is no phase information of any kind for this variant: the case holds no proband genotype, no family structure, no long-read or allele-specific phasing, and no co-occurrence data with a pathogenic RAD51C allele. The four ClinVar submissions for VCV000492371 are single-submitter Likely benign records without expert-panel review, and the submission audit found no criterion lead asserting a cis or trans relationship with a pathogenic RAD51C variant. No retrieved publication mentions the variant at all (the only two full texts available, PMID:25741868 and PMID:25394175, never name RAD51C), so no cis/trans configuration is reported in the literature either; the Fanconi Anemia GeneReviews chapter (PMID:20301575) and Sherloc (PMID:28492532) were not retrievable as text for this variant. Population data were examined and do not substitute for the required observation: gnomAD v4.1 lists 26 alleles for this variant among 1,613,178 (AF 1.61e-05, grpmax FAF 1.377e-05) with a single homozygote in European (non-Finnish) individuals, and gnomAD v2.1 lists 4 of 251,368 alleles with 0 homozygotes. A homozygote carries the variant on both homologs - it is the variant paired with itself, not a documented cis or trans relationship with a pathogenic variant - so it falls outside BP2's definition and cannot be used to claim BP2 support. With no cis/trans-with-pathogenic observation available, BP2 is not met; no strength tier is assigned. No RAD51C-specific modification exists because the nominally governing ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer RAD51C specification v1.0 (doc 1535377178) exposes an empty criterion-level rule payload, so the generic ACMG/AMP 2015 BP2 definition was applied.

Rule

ACMG/AMP 2015 BP2 (PMID:25741868): observed in trans with a pathogenic variant for a fully penetrant dominant gene/disorder, or observed in cis with a pathogenic variant in any inheritance pattern. No RAD51C-specific VCEP rule text available (CSPEC v1.0 doc 1535377178 rule payload empty). Verdict: not met - no phased cis/trans relationship with a pathogenic RAD51C variant is documented for this variant, and population homozygosity is not a BP2 observation.

Evidence considered

• BP2 definition applied (ACMG/AMP 2015, PMID:25741868): observed in trans with a pathogenic variant for a fully penetrant dominant gene/disorder, or observed in cis with a pathogenic variant in any inheritance pattern. Both arms require a phased observation relative to a pathogenic variant. The retrieved copy of PMID:25741868 is truncated at 40,000 characters and ends before Table 3, so the BP2 criterion row wording could not be quoted from the retrieved copy; the definition used is the standard ACMG/AMP 2015 criterion text, and the retrieved copy's verbatim statement on PM3 confirms that in-trans detection relative to a pathogenic variant is treated by this framework as an observed, case-level finding rather than something inferable from frequency data or from another laboratory's classification.
• Inheritance-mode context: RAD51C is associated with autosomal dominant familial ovarian cancer and hereditary breast carcinoma and with autosomal recessive Fanconi anemia (MONDO:0019391), so both BP2 arms were considered applicable to the gene rather than dismissed on inheritance grounds.
• Variant identity: NM_058216.3:c.571+15G>T (NC_000017.10:g.56774235G>T; NC_000017.11:g.58696874G>T), an intronic substitution 15 nucleotides into intron 3 (variant_exonic_positions 3i) with predicted protein consequence p.? - a single-variant record with no accompanying second variant, so no in-case allele pairing exists.
• No proband, phase, cis/trans, compound-heterozygous, or co-occurrence content is present anywhere in the case evidence for this variant.
• ClinVar VCV000492371 lists four single-submitter Likely benign records (GeneDx, Color Diagnostics, Labcorp/Invitae, Sinai Health System), no expert panel; the submission audit extracted no criterion lead of the cis/trans type, so ClinVar provides no phasing evidence for BP2. The final ClinVar classification was used only as context and as a lead-screening signal, never as an inference for BP2.
• gnomAD v4.1 (GRCh38, chr17-58696874-G-T): 26 of 1,613,178 alleles (AF 1.61173e-05, MAF 0.00161%), total homozygotes 1; highest-frequency population European (non-Finnish) with 24 of 1,179,318 alleles (AF 2.03507e-05) and 1 homozygote; grpmax FAF 1.377e-05. The homozygote means the variant is present on both homologs of one individual (variant paired with itself), not a cis or trans relationship with a pathogenic RAD51C variant.
• gnomAD v2.1 (GRCh37, 17-56774235-G-T): 4 of 251,368 exome alleles (AF 1.59129e-05, MAF 0.00159%), 0 homozygotes; highest single-population AF 0.000163132 (1/6,130 alleles, Remaining individuals), grpmax FAF 7.01e-06.
• Population frequencies were reviewed only to test whether any homozygous or co-occurrence observation could inform BP2; they carry no BP2-compatible cis/trans information and were not used as a frequency criterion.
• Literature scope: PMID:25741868 (ACMG/AMP framework) and PMID:25394175 (ACMG/NSGC referral guideline) were retrieved in full and neither names RAD51C or this variant; PMID:20301575 (Fanconi Anemia GeneReviews) has only a screenshot in this case and PMID:28492532 (Sherloc) has no full text or extract, so neither could be screened for a cis/trans observation.
• Governing framework check: the ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer Expert Panel specification for RAD51C v1.0 (CSPEC doc 1535377178, source key 'cspec') is present but exposes no criterion-level rule payload, so the generic ACMG/AMP 2015 BP2 definition was applied without gene-specific modification.

Still missing

• No phasing data (long-read sequencing, allele-specific assays, or informative family genotypes) exist to place c.571+15G>T in cis or in trans with any RAD51C variant, pathogenic or otherwise; BP2 would become assessable only with such data.
• The single gnomAD v4.1 homozygote (European non-Finnish) has no accompanying phenotype or health-status information, so it cannot be developed into healthy-homozygote evidence either; that consideration belongs to the benign-observation criteria, not to BP2.
• The ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer RAD51C specification (v1.0, doc 1535377178) contains no retrievable criterion-level BP2 rule text, so no gene-specific BP2 restatement or strength modification could be applied.
• PMID:20301575 (Fanconi Anemia, GeneReviews) and PMID:28492532 (Sherloc) were not available as full text or pre-computed extraction, so they could not be screened for cis/trans observations involving this variant.

Sources

Reasoning

BP6 requires an exact-variant expert-panel assertion of Benign or Likely benign: an aggregate ClinVar label or collection of ordinary laboratory submissions must never trigger it. The exact-variant record here is VCV000492371 (NM_058216.3(RAD51C):c.571+15G>T, exact match by transcript HGVS), whose summary is Likely benign from four clinical laboratories - but the provenance is wrong for BP6 on every element. All four SCVs are ordinary clinical-testing submissions from testing laboratories (GeneDx, Color Diagnostics/Color Health, Labcorp Genetics/Invitae, Sinai Health) and the submission audit marks is_expert_panel = false on each, with expert_panel_submissions = 0. The ClinVar review status is 'criteria provided, multiple submitters, no conflicts' (2 stars), which is distinct from 'reviewed by expert panel' and therefore encodes no expert-panel judgement. Consequently, the benign signal in ClinVar cannot be scored through BP6: what exists is four independent non-expert Likely benign assertions and an aggregate label, which the governing rule for this group expressly excludes, and no criterion was inferred from any submitter's final classification. The HBOP/RAD51C VCEP specification v1.0 (CSPEC doc 1535377178) is the governing framework and is present, but its criteria payload is empty, so it supplies no gene-specific BP6 rule that could change this. BP6 is therefore not met; any benign weighting from this variant's population or computational evidence must come from the criteria that own that evidence, not from an expert-panel-assertion criterion whose precondition is absent.

Rule

BP6 (ACMG/AMP; Richards et al. 2015, PMID:25741868): a reputable source asserts the variant is benign - operationalised for this group as an exact-variant ClinGen expert-panel Benign or Likely benign classification only. Aggregate ClinVar labels and non-expert submissions never trigger BP6.

Evidence considered

• Exact-variant ClinVar record VCV000492371 for NM_058216.3(RAD51C):c.571+15G>T: aggregate classification Likely benign (4 clinical laboratories), review status 'criteria provided, multiple submitters, no conflicts' (2 stars).
• Submission audit for the same record: submissions_total = 4, expert_panel_submissions = 0, and is_expert_panel = false for SCV000729555 (GeneDx), SCV000691247 (Color Diagnostics, LLC DBA Color Health), SCV002397711 (Labcorp Genetics, formerly Invitae) and SCV005918798 (Department of Pathology and Laboratory Medicine, Sinai Health System).
• Under the rule governing this group, an exact-variant expert-panel Benign/Likely benign classification is the only trigger for BP6; ordinary laboratory submissions, aggregate ClinVar labels and non-expert assertions must never trigger it, and that precondition is not satisfied here.
• The HBOP VCEP RAD51C specification v1.0 (CSPEC doc 1535377178) is the governing framework and is present in the case, but its criteria/rules payload is empty, so no gene-specific BP6 rule modifies the general requirement.
• No criterion was inferred from any submitter's final classification, and no numeric threshold or comparison direction is attached to BP6, so no inequality was constructed.

Still missing

• No ClinGen expert-panel (3-star) Benign or Likely benign classification exists for RAD51C c.571+15G>T; BP6 would need re-evaluation if the HBOP VCEP ever deposits such an assertion.

Sources

Reasoning

PS2 requires a de novo occurrence of the variant in a proband with the relevant phenotype, with both maternity and paternity confirmed (Richards et al. 2015, PMID:25741868; generic ACMG/AMP, as the ClinGen HBOP CSPEC RAD51C v1.0 ruleset supplied for this case carries no criterion-level specifications for PS2 and was retrieved only as an unstructured rule origin). This case record contains no proband and no parental/trio testing at all: the evidence is variant-level curation only (population frequency, in-silico splicing, ClinVar submissions). All four ClinVar submissions on variation 492371 (SCV000729555 GeneDx, SCV000691247 Color Diagnostics, SCV002397711 Labcorp/Invitae, SCV005918798 Sinai Health System) are germline, 'criteria provided, single submitter' likely-benign entries, and the submission audit recovers no de novo claim and no parental-testing result for any of them (the only criterion signal recovered is a PP3 in-silico signal in SCV000729555). No pedigree, no family history, and no parental genotypes exist in any retrieved source, so a confirmed de novo occurrence can be neither demonstrated nor excluded. The criterion is therefore not assessed; the gap is a proband-level record with confirmed trio testing.

Evidence considered

• Variant under assessment is NM_058216.3:c.571+15G>T (NC_000017.10:g.56774235G>T / NC_000017.11:g.58696874G>T), an intronic change annotated by VariantValidator as intron 3 (variant_exonic_positions start_exon/end_exon = 3i) with no predicted protein consequence (NP_478123.1:p.?); it is not a coding or splice-site change.
• The RAD51C CSPEC document (ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer Expert Panel Specifications, v1.0, doc 1535377178) was retrieved as an unstructured ruleset whose rule_payload is empty and whose criteria list is empty; no PS2/PM6/PP1/BS4 specification, strength modification, or phenotype-specificity requirement could be read from it. Its only captured content is gene/disease context: RAD51C is linked to familial ovarian cancer and hereditary breast carcinoma (autosomal dominant) and Fanconi anemia (autosomal recessive).
• No proband record exists in this case: neither case_summary.json, evidence.json, evidence_brief.json, literature_pass.json, fulltext_map.json, paper_extracts.json, prefetch.json nor vcep_materials.json contains a patient, phenotype, HPO term, age of onset, or family-history field for this variant.
• ClinVar variation 492371 carries four germline submissions, all classified Likely benign with review status 'criteria provided, single submitter' and no expert-panel input: SCV000729555 (GeneDx), SCV000691247 (Color Diagnostics), SCV002397711 (Labcorp Genetics/Invitae), SCV005918798 (Sinai Health System). The submission audit recovered only a single in-silico PP3 signal (SCV000729555); none of the four records reports a de novo occurrence, parental testing, or family genotyping.
• ClinVar citation trail from this variant leads to PMID:25741868, PMID:28492532, PMID:25394175 and PMID:20301575; none of the retrieved extracts or abstracts reports a proband, trio, or de novo event for RAD51C c.571+15G>T.

Still missing

• No proband-level clinical record (phenotype, age at diagnosis, family history) for this variant in the case evidence.
• No parental/trio genotyping confirming maternity and paternity (e.g. STR or SNP panel) was performed or reported.
• No de novo occurrence observed in any retrieved source; the four ClinVar submissions carry no de novo or parental-testing information.
• Gene-specific PS2 strength specification is unavailable: the RAD51C CSPEC v1.0 ruleset was retrieved with an empty rule payload, so generic ACMG/AMP 2015 PS2 is the only applicable rule.

Sources

Reasoning

PS3 requires well-established functional studies showing a damaging effect on the gene or gene product. NM_058216.3(RAD51C):c.571+15G>T is a deep-intronic change 15 nucleotides into intron 3 (variant validator places it in exon 3i; SpliceAI max delta 0.006, i.e. no predicted splice impact), and per ACMG/AMP 2015 (Richards et al., PMID:25741868) the relevant functional evidence for such a splice-region variant would be an mRNA (RNA) or protein assay demonstrating aberrant splicing or loss of function. No such assay exists in the retrieved evidence: no RNA/minigene/RT-PCR study, no protein- or cell-based assay (e.g. MMC sensitivity, RAD51 focus formation) has been performed on or reported for this specific variant. The only RAD51C functional work encountered in the case literature is gene-level (Vaz et al. 2010, PMID:20400963, RAD51C focus formation and mitomycin C hypersensitivity in a Fanconi anemia-like family carrying a different homozygous RAD51C missense variant), which supports RAD51C loss of function as a disease mechanism generally but says nothing about c.571+15G>T. ClinVar submissions for this variant (ClinVar 492371, four clinical laboratories, all Likely benign) cite only in-silico reasoning (one submitter's criteria_found = PP3), no functional criterion. Because no qualifying functional study was identified, PS3 is not assessed rather than not met, and the reviewable gap is the absence of an RNA-based splice assay.

Rule

ACMG/AMP 2015 PS3: well-established functional studies show a damaging effect on the gene or gene product (Richards et al., PMID:25741868); splice-region variants require RNA or protein functional confirmation.

Evidence considered

• ACMG/AMP 2015 (Richards et al., PMID:25741868) defines PS3 as well-established functional studies showing a damaging effect, and explicitly states that for splice-site variants, confirmation of impact requires functional analysis by either RNA or protein analysis; mRNA-level assays are described as informative for variants at splice junctions and deeper intronic regions.
• Variant context: NM_058216.3:c.571+15G>T (NC_000017.10:g.56774235G>T, GRCh38 g.58696874G>T), a non-canonical deep-intronic change 15 nt into intron 3 with predicted protein consequence NP_478123.1:p.?.
• SpliceAI max delta score = 0.006 (DS_AG 0.003, DS_AL 0.000, DS_DG 0.003, DS_DL 0.006; Pangolin SG 0.01, SL -0.006). This is an in-silico prediction, not a functional assay, and per ACMG/AMP 2015 (PMID:25741868) in-silico results cannot serve as the sole evidence for an assertion; it therefore contributes nothing to PS3.
• ClinVar variation 492371 submissions (GeneDx SCV000729555, Color Health SCV000691247, Labcorp/Invitae SCV002397711, Sinai Health System SCV005918798): all Likely benign, no expert-panel submission, no functional criterion cited; the only criterion signal recovered was PP3 (in-silico) in SCV000729555.
• Gene-level context reviewed and found non-variant-specific: Vaz et al. 2010 (PMID:20400963) reported loss of RAD51 focus formation and increased mitomycin C/camptothecin sensitivity with a homozygous RAD51C missense variant in a Fanconi anemia-like family, establishing a RAD51C loss-of-function disease mechanism but not testing c.571+15G>T.
• No RAD51C-specific PS3 specification was available from the governing framework: the CSPEC entry for RAD51C (doc 1535377178, 'Hereditary Breast, Ovarian and Pancreatic Cancer Specification v1.0') was retrieved but its criteria list was empty and no local RAD51C VCEP files were present, so generic ACMG/AMP PS3 definitions were applied.

Still missing

• No RNA (RT-PCR/minigene) or protein functional assay has been reported for RAD51C c.571+15G>T, so the splice-region effect central to this variant remains untested experimentally.
• A targeted search of RAD51C splice-variant functional literature (beyond the PMIDs already retrieved for this case) would be required to close this gap fully; the four surviving ClinVar-linked PMIDs and the gene-level RAD51C papers contain no variant-specific assay.
• No RAD51C-specific VCEP PS3 rule was retrievable to calibrate strength even if an assay existed (HBOP VCEP RAD51C v1.0 CSPEC document 1535377178 was located but its criteria payload was empty), so any future assay would default to generic ACMG PS3.

Sources

Reasoning

PS4 (Richards et al. 2015, PMID:25741868) requires case-control enrichment (a statistically elevated variant frequency in affected cases versus controls) or, in looser implementations, at least two independent probands with the variant whose phenotype matches the gene. The governing framework was checked first: the ClinGen Hereditary Breast, Ovarian and Pancreatic Cancer (HBOP) VCEP specification for RAD51C v1.0 (CSPEC doc 1535377178) is present in this case but carries an empty criteria/rules payload (framework_complete=false, rule_payload={}), so no gene-specific PS4 definition, odds-ratio cutoff or operator is available and generic ACMG/AMP PS4 governs; because no numeric PS4 value exists for this variant, no inequality was written or evaluated, and no operator was pattern-matched from any other criterion in this group. Evidence search for this exact variant: the ClinVar record VCV000492371 (NM_058216.3(RAD51C):c.571+15G>T, exact transcript-HGVS match) carries four germline clinical-testing submissions, all Likely benign, with no case-level detail, no proband counts, no phenotype descriptions and no expert-panel submission; its cited PMIDs were screened specifically for PS4 (PMID:20301575, PMID:25394175, PMID:25741868, PMID:28492532) and none reports the variant, an affected proband, a case series or any enrichment statistic — PMID:25394175 and PMID:25741868 were read in full and never name RAD51C, PMID:28492532 is a criteria-refinement methods paper without variant-level data, and PMID:20301575 (GeneReviews: Fanconi Anemia) was not retrievable as full text. A supplementary PubMed search for RAD51C c.571 variants identified PMID:33333735 (BRIDGES: 60,466 breast cancer cases, 53,461 controls), which functionally and clinically characterised 20 RAD51C splice-region variants and reported c.571+4A>G and c.571+5G>A as pathogenic/likely pathogenic; c.571+15G>T was not among them, so that cohort supplies no enrichment signal for this variant either. The population frequencies on record (gnomAD v2.1 AF 1.59e-05, 4/251,368 alleles; gnomAD v4.1 AF 1.61e-05, 26/1,613,178 alleles with 1 homozygote) are allele-frequency evidence that belongs to the population group and cannot be converted into PS4 case-control enrichment. With zero affected probands and zero case-control datasets, PS4 cannot be evaluated at all for this variant; the honest verdict is not_assessed rather than not_met, and the gap is the absence of any case or control data.

Rule

PS4 (Richards et al. 2015, PMID:25741868): case-control enrichment (elevated odds ratio / relative risk in affected cases) or >= 2 independent probands with the phenotype. No gene-specific RAD51C/HBOP VCEP PS4 rule or threshold was retrievable, so generic ACMG/AMP PS4 applies.

Evidence considered

• ClinVar record VCV000492371 (NM_058216.3(RAD51C):c.571+15G>T, variation ID 492371) is an exact transcript-HGVS match: four germline clinical-testing submissions (GeneDx SCV000729555; Color Health SCV000691247; Labcorp/Invitae SCV002397711; Sinai Health SCV005918798), all Likely benign, each 'criteria provided, single submitter', with no proband, family, phenotype, segregation or case-control data recorded.
• The submission audit records expert_panel_submissions = 0 and no case-level evidence; the ClinVar conditions attached to the record (breast-ovarian cancer, familial, susceptibility to, 3; Fanconi anemia complementation group O; hereditary cancer-predisposing syndrome; not specified) are record-level condition labels, not a case series usable as PS4 enrichment.
• The four ClinVar-derived PMIDs were screened specifically for PS4: none reports RAD51C c.571+15G>T, an affected carrier, a case series or a case-control analysis.
• PMID:25394175 (Hampel et al., ACMG/NSGC referral guideline) was read in full: it is a cancer-predisposition referral guideline, does not name RAD51C, does not mention the variant, and presents no cohort, functional, segregation or case data.
• PMID:25741868 (Richards et al. 2015, ACMG/AMP standards) is a general classification framework: it defines PS4 as case-control enrichment or multiple independent affected probands, but reports no RAD51C case or cohort data and never names RAD51C.
• PMID:28492532 (Nykamp et al., Sherloc) is a criteria-refinement methods paper (40,000+ clinically observed variants used to refine ACMG/AMP criteria); its abstract contains no variant-level data and does not mention c.571+15G>T.
• PMID:20301575 (GeneReviews: Fanconi Anemia) was cited from the ClinVar prefetch and triaged for PS4/PP5 but no full text was fetched for it, so it contributed nothing verifiable; it was not treated as evidence either way.
• Supplementary PubMed search (NCBI E-utilities, queried for RAD51C c.571 variants and for the genomic coordinates g.56774235 / chr17:56774235) returned no publication reporting this variant. The nearest cohort resource is PMID:33333735 (Sanoguera-Miralles et al., Cancers 2020, BRIDGES project), which sequenced RAD51C in 60,466 breast cancer patients and 53,461 controls and functionally assayed 20 splice-region variants; it reports c.571+4A>G and c.571+5G>A as pathogenic/likely pathogenic but not c.571+15G>T.
• No numeric PS4 threshold or operator was available from the governing VCEP specification or from a named publication for this variant, so no inequality such as 'value >= threshold' or 'value <= threshold' was constructed for PS4 and none was borrowed from another criterion in this group.
• gnomAD data on record (v2.1 AF 1.59129e-05, 4/251,368 alleles; v4.1 AF 1.61173e-05, 26/1,613,178 alleles, 1 homozygote; grpmax FAF 1.377e-05) is population allele-frequency evidence, not case-control enrichment, and was not used as PS4 support.

Still missing

• No case-control cohort, case series or individual proband carrying RAD51C c.571+15G>T was identified in the case evidence or in a supplementary PubMed search, so no odds ratio or equivalent enrichment statistic exists for this variant and PS4 cannot be evaluated.
• PMID:20301575 (GeneReviews: Fanconi Anemia), cited from the ClinVar prefetch and triaged for PS4/PP5, was never retrieved as full text (only a screenshot page exists in the case), so no variant-level or case-level content in it could be checked.
• No gene-specific PS4 rule (definition, threshold or operator) is present in the HBOP/RAD51C VCEP specification payload, so generic ACMG/AMP PS4 was applied as the fallback framework.

Sources

Reasoning

PM6 applies to an assumed de novo occurrence without confirmation of maternity and paternity (Richards et al. 2015, PMID:25741868; generic ACMG/AMP, since the RAD51C CSPEC v1.0 ruleset for this case contains no criterion-level PM6 specification). Like PS2, PM6 still requires an observed de novo event in a proband with the relevant phenotype; the lower bar concerns only the confirmation of parentage, not the existence of the proband or of the parental testing. This case has no proband, no phenotype, and no parental testing of any kind, confirmed or unconfirmed. The four ClinVar submissions on variation 492371 are germline, single-submitter likely-benign records and report no de novo event and no parental genotypes. With neither a de novo observation nor any parental data, PM6 is not assessable.

Evidence considered

• PM6 was evaluated using the generic ACMG/AMP 2015 definition because no RAD51C CSPEC v1.0 criterion-level text for PM6 was recoverable (retrieved ruleset has an empty rule_payload and an empty criteria list).
• No proband-level clinical record, phenotype, or family history for this variant appears anywhere in the case evidence; the case is limited to variant-level data sources (gnomAD v2.1/v4.1, SpliceAI, ClinVar, COSMIC, OncoKB, REVEL/BayesDel lookups).
• All four ClinVar submissions for variation 492371 are germline clinical-testing submissions with no reported parental genotyping and no de novo statement; the audit's only recovered criterion signal is an in-silico PP3 in SCV000729555.
• RAD51C c.571+15G>T is an intronic (intron 3) change with p.? protein consequence and SpliceAI max delta 0.006, i.e. no predicted transcript-level consequence that would itself imply a de novo functional event.

Still missing

• No proband with the relevant RAD51C-associated phenotype (ovarian/breast cancer or Fanconi anemia) is documented for this variant.
• No parental genotypes (even without confirmation of parentage) are available to support an assumed de novo event.
• No source in this case reports a de novo occurrence of RAD51C c.571+15G>T.
• Gene-specific PM6 strength and applicability rules could not be recovered from the RAD51C CSPEC v1.0 ruleset (empty rule payload).

Sources

Reasoning

PP1 requires co-segregation of the variant with disease in multiple affected family members (Richards et al. 2015, PMID:25741868); the generic ACMG/AMP definition was used because the RAD51C CSPEC v1.0 ruleset for this case exposes no criterion-level PP1 text. Segregation evidence is entirely absent here: no pedigree, no family members, no meiosis count, and no affected/unaffected genotype data exist in any retrieved source, and no publication retrieved for this variant reports a family. The four ClinVar submissions on variation 492371 report no family testing and provide no segregation claim. Because zero informative meioses are available, neither co-segregation nor the absence of it can be observed, so PP1 cannot be assessed; the gap is family genotyping data.

Evidence considered

• PP1 is defined in the generic ACMG/AMP 2015 framework as co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease, with strength modifiable by the number of informative meioses (PMID:25741868); that framework paper contains no RAD51C-specific or variant-specific segregation data.
• The full-text extraction for PMID:25394175 confirms it is an ACMG/NSGC referral-indications practice guideline that names no RAD51C variant and reports no segregation, family, or variant-level data for RAD51C c.571+15G>T.
• Literature retrieval for this case returned only four PMIDs (25741868, 28492532, 25394175, 20301575), all ClinVar citation leads; none is a RAD51C family or segregation study, and the case's criteria_index maps only PP5 and PS4 candidate criteria, with no segregation-themed paper retrieved.
• ClinVar variation 492371 (four germline, single-submitter Likely benign records) contains no family testing, no pedigree, and no co-segregation statement.
• No pedigree, proband, or relative genotype field exists in any case file, so the number of informative meioses cannot even be counted.

Still missing

• No pedigree or family structure for the variant is documented in any retrieved source.
• No genotype data for affected or unaffected relatives; the number of informative meioses is zero.
• No published family showing co-segregation of RAD51C c.571+15G>T with ovarian/breast cancer or Fanconi anemia was retrieved.
• Phenotype definition for the family (which RAD51C-associated disease to segregate with) is unspecified, and the CSPEC v1.0 ruleset supplied no gene-specific PP1 rule.

Sources

Reasoning

PP4 (Richards et al. 2015, PMID:25741868) requires that the affected individual's phenotype or family history be highly specific for a single gene with a single known genetic aetiology, such that the clinical presentation itself constrains the gene-level diagnosis. This case is a variant-level classification: neither the case summary nor any other case file carries HPO terms, a patient description, age at diagnosis, tumour type, family history, or any other proband-level clinical data, and no affected carrier of c.571+15G>T is described anywhere in the evidence. The only phenotype strings present are the ClinVar record-level conditions attached to variant 492371 (breast-ovarian cancer, familial, susceptibility to, 3; Fanconi anemia complementation group O; hereditary cancer-predisposing syndrome; not specified); these are annotations of the variant record rather than of an individual patient, so they cannot establish that any proband's presentation is gene-specific. Those labels also cut against specificity on their face: they span the autosomal-dominant hereditary breast/ovarian cancer context and the autosomal-recessive Fanconi anemia context for the same gene, and HBOC is a genetically heterogeneous presentation shared with BRCA1/BRCA2/PALB2/RAD51D and others, so a stated HBOC-type phenotype would not by itself be highly specific for RAD51C. No numeric threshold is attached to this criterion, so no inequality or comparison direction was applied and none was inferred from any other criterion in this group. Because PP4 is entirely proband-phenotype driven and no proband phenotype exists in the case evidence, the criterion cannot be evaluated; the honest verdict is not_assessed, with the missing phenotype/family-history data recorded as the gap.

Rule

PP4 (Richards et al. 2015, PMID:25741868): the individual's phenotype or family history must be highly specific for a single gene with a single known genetic aetiology; requires proband-level phenotype data, which is absent here. No gene-specific RAD51C/HBOP VCEP PP4 rule was retrievable.

Evidence considered

• PP4 definition (Richards et al. 2015, PMID:25741868): the variant is found in a patient whose phenotype or family history is highly specific for a single gene with a single known genetic aetiology.
• The case files (case summary, review manifest, evidence bundle) contain no HPO terms, no patient narrative, no family history and no affected-carrier description; the classification is variant-level only.
• ClinVar record VCV000492371 lists these condition labels for the variant: 'Breast-ovarian cancer, familial, susceptibility to, 3', 'Fanconi anemia complementation group O', 'Hereditary cancer-predisposing syndrome', 'not specified' - all attached to the variant record, not to an individual patient.
• RAD51C disease context from the HBOP VCEP RAD51C specification (CSPEC doc 1535377178) raw ruleset: familial ovarian cancer (AD, MONDO:0016248), hereditary breast carcinoma (AD, MONDO:0016419), Fanconi anemia (AR, MONDO:0019391) - i.e. two distinct inheritance modes and a genetically heterogeneous dominant presentation, which is the opposite of single-gene-specific phenotyping.
• No numeric PP4 threshold or operator exists in the governing specification or in any named publication for this criterion, so no inequality comparison was performed.

Still missing

• No individual-level clinical data (phenotype, HPO terms, age at diagnosis, tumour type, family history, cascade testing) is present in the case evidence for the carrier of RAD51C c.571+15G>T, so phenotype specificity for a single gene cannot be assessed.
• The only phenotype labels in the case are ClinVar record-level conditions for the variant (breast-ovarian cancer familial susceptibility 3, Fanconi anemia complementation group O, hereditary cancer-predisposing syndrome, not specified); they are not proband phenotypes and they span both the dominant HBOC and recessive Fanconi anemia contexts for RAD51C.

Sources

Reasoning

BS3 requires well-established functional studies showing no damaging effect on protein function or splicing. For the deep-intronic RAD51C c.571+15G>T (15 nt into intron 3), the decisive evidence would be an RNA assay demonstrating normal transcript processing or a protein/cell assay showing preserved RAD51C function. No such assay was identified for this variant in any retrieved source: the case literature contains no RNA, minigene, protein, or cell-based functional analysis of c.571+15G>T, and the ClinVar record (492371, four clinical laboratories, all Likely benign) rests on in-silico reasoning only (one submitter's criterion signal was PP3), which per ACMG/AMP 2015 (Richards et al., PMID:25741868) cannot be used as functional evidence. The benign-direction argument for this variant therefore cannot be anchored to a functional assay; BS3 is left not assessed, with the gap being the absence of an assay demonstrating no splice/functional impact.

Rule

ACMG/AMP 2015 BS3: well-established functional studies show no damaging effect on protein function or splicing (Richards et al., PMID:25741868).

Evidence considered

• ACMG/AMP 2015 (Richards et al., PMID:25741868) defines BS3 as well-established functional studies showing no damaging effect on protein function or splicing; the same framework notes that in-silico splice predictions cannot substitute for functional analysis and that a splice-region variant's impact requires RNA or protein confirmation.
• No variant-specific functional assay for RAD51C c.571+15G>T was found in the retrieved sources; the only RAD51C functional evidence located (Vaz et al. 2010, PMID:20400963: loss of RAD51 focus formation and mitomycin C hypersensitivity) concerns a different, homozygous RAD51C missense variant in a Fanconi anemia-like family and is gene-level only.
• SpliceAI max delta score = 0.006 for this variant is a computational prediction of no significant splice impact; because it is in-silico it is treated as part of the PP3/BP4 predictor evidence stream (assessed elsewhere) and is not accepted as BS3 functional evidence, per ACMG/AMP 2015 (PMID:25741868).
• ClinVar variation 492371: four Likely benign submissions, none from an expert panel, none citing a functional criterion; criterion signal recovered was PP3 (in-silico) only, so the ClinVar record provides no BS3 basis.
• Governing framework check: RAD51C CSPEC document 1535377178 ('Hereditary Breast, Ovarian and Pancreatic Cancer Specification v1.0') retrieved with an empty criteria set and no local RAD51C VCEP specification files present; therefore no gene-specific BS3 rule was available and generic ACMG/AMP applied.

Still missing

• No functional assay demonstrating a normal (non-damaging) effect exists for RAD51C c.571+15G>T, so the BS3 direction is unsupported by assay data.
• An RNA-based splice assay (RT-PCR of patient transcript or minigene) showing normal exon 4 inclusion/transcript processing would be needed to convert this to a usable BS3; none has been reported in the retrieved literature.
• No RAD51C-specific VCEP BS3 rule was retrievable for strength calibration (HBOP VCEP RAD51C v1.0 CSPEC criteria payload empty; no local RAD51C VCEP files), so generic ACMG/AMP BS3 definitions would apply.

Sources

Reasoning

BS4 requires lack of segregation in affected members of a family (Richards et al. 2015, PMID:25741868); the generic ACMG/AMP rule was used because the RAD51C CSPEC v1.0 ruleset for this case provides no criterion-level BS4 specification. Non-segregation is an observation about family genotypes, and no family data exist in this case: there is no pedigree, no affected relative, no unaffected relative, and no meiosis count in any retrieved source, and the four ClinVar submissions on variation 492371 report no family testing. The gnomAD frequency data (v2.1 AF 1.59e-05, 4/251368 alleles; v4.1 AF 1.61e-05, 26/1613178 alleles with 1 homozygote) and the ClinVar likely-benign consensus are population- and submitter-level evidence that belong to the population-frequency and criteria-assertion groups, not to BS4, and cannot be converted into a non-segregation observation. BS4 is therefore not assessed rather than met.

Evidence considered

• BS4 is defined in the generic ACMG/AMP 2015 framework as lack of segregation in affected members of a family (Richards et al. 2015, PMID:25741868); the framework paper provides the definition only and contains no RAD51C data.
• No familial non-segregation evidence exists in this case: none of the four ClinVar submissions (SCV000729555, SCV000691247, SCV002397711, SCV005918798) reports family testing, and no pedigree or relative genotype data appears in any case file.
• The retrieved population-frequency data (gnomAD v2.1 AF 1.59129e-05 with 4/251368 alleles and 0 homozygotes; gnomAD v4.1 AF 1.61173e-05 with 26/1613178 alleles and 1 homozygote) is allele-frequency evidence relevant to benign frequency criteria, not to BS4, which requires genotype data from relatives.
• The four independent ClinVar Likely-benign submissions for this variant are submitter classifications and were not treated as a substitute for family segregation data; per the case instructions, no criterion may be inferred from another source's final classification.
• RAD51C c.571+15G>T is a deep-intronic (intron 3) variant with no predicted protein consequence (NP_478123.1:p.?), so no transcript-level observation is available that could substitute for genotype-based non-segregation.

Still missing

• No family (affected or unaffected relatives) is genotyped, so lack of segregation cannot be observed.
• No pedigree or family history for this variant in any retrieved source.
• No publication reporting non-segregation of RAD51C c.571+15G>T was retrieved.
• Gene-specific BS4 rule was unavailable from the RAD51C CSPEC v1.0 ruleset (empty rule payload).

Sources

Reasoning

BP5 in generic ACMG/AMP (Richards et al. 2015, PMID:25741868) reads 'variant found in a case with an alternate molecular basis for disease' - it is a case-level criterion that requires a fully characterised affected individual in whom a different variant or gene already explains the phenotype. This case is a variant-level classification: there is no proband in the evidence, no genotype beyond the variant itself, no information about a second pathogenic variant (in RAD51C or elsewhere), no tumour or family testing, and no clinical narrative. The ClinVar record for the variant adds nothing case-level: its four submissions are germline clinical-testing entries with no conditions or case description on three of them and only the label 'not specified' on the fourth, and no case details are provided. A framework-specific version of BP5 could in principle be defined numerically (some VCEP specifications formulate BP5 as a likelihood-ratio-style quantity that strengthens as it decreases, met at 'value <= threshold'), but no such formulation is available here: the governing HBOP/RAD51C VCEP specification v1.0 (CSPEC doc 1535377178) has an empty criteria/rules payload, so no gene-specific BP5 operator or cutoff exists for RAD51C. Because no numeric BP5 value and no BP5 operator were available, no inequality comparison was written for BP5, and the operator was deliberately not inferred from any other criterion in this group; with no case-level alternate-basis data at all, the criterion cannot be evaluated and is recorded as not_assessed rather than not_met.

Rule

BP5 (Richards et al. 2015, PMID:25741868): variant found in a case with an alternate molecular basis for disease. No gene-specific RAD51C/HBOP VCEP BP5 rule, value or operator was retrievable, so no numeric threshold was applied.

Evidence considered

• BP5 definition (Richards et al. 2015, PMID:25741868): variant found in a case with an alternate molecular basis for disease - a case-level requirement.
• This classification is variant-level: no proband, no second-variant or co-occurring variant information, no tumour or family testing, and no clinical narrative are present anywhere in the case evidence.
• ClinVar record VCV000492371 provides no case-level detail: four germline clinical-testing submissions, three with empty condition_labels and one with 'not specified', and no case description in the submission audit.
• The HBOP VCEP RAD51C specification v1.0 (CSPEC doc 1535377178) was consulted for a gene-specific BP5 rule and has an empty criteria/rules payload (framework_complete=false, rule_payload={}), so no BP5 threshold, value or comparison operator exists to apply.
• Because neither a numeric BP5 value nor a BP5-specific operator was available, no inequality (for example 'value <= threshold') was written or evaluated, and no operator was pattern-matched from PS4 or any other criterion in this group.

Still missing

• No proband-level data (second variant, co-occurring pathogenic variant, tumour testing, family testing) exists in the case evidence, so an alternate molecular basis for disease cannot be identified or excluded.
• No gene-specific BP5 formulation (including any likelihood-ratio-style version) is present in the HBOP/RAD51C VCEP specification payload, so no BP5 threshold or operator was available and none was inferred from another criterion.

Sources