RAD51B c.585_587dup, p.(Glu198dup)

Reasoning

PM2 (absent or extremely low frequency in population databases) is met at supporting strength. The variant is present at extremely low frequency: gnomAD v2.1 total AF 9.09695e-06 (0.00091%, 2/219,854 alleles) with max subpopulation AF 0.00190% (NFE); gnomAD v4.1 total AF 1.99861e-06 (0.00020%, 3/1,501,040 alleles) with max NFE AF 0.00027%; grpmax FAF 3.15e-06 (v2.1) and 7.3e-07 (v4.1); zero homozygotes in both releases; absent from gnomAD-Canada v1.0. All frequencies are well below the generic non-VCEP PM2 threshold of <0.1%, consistent across two independent gnomAD releases and an additional Canadian genome cohort. PM2 is applied at supporting strength per the SVI-calibrated generic framework (Tavtigian et al. 2018, PMID 28492532), rather than the original moderate assignment. Threshold artifacts considered: data are exome-only in both releases (genome AC=0/AN=0), and the 3-bp in-frame duplication is counted consistently (AC=2 v2.1, AC=3 v4.1), so no under-ascertainment or alignment artifact is suspected; the low allele counts nevertheless cap confidence at moderate.

Rule

Generic ACMG/AMP 2015 (PMID 25741868) PM2: absent or at extremely low frequency in population databases (<0.1% local non-VCEP threshold). Met; strength supporting (PMID 28492532).

Evidence considered

• gnomAD v2.1 exome: total AF 9.09695e-06 (0.00091%, 2/219854 alleles), max NFE AF 1.89836e-05 (0.00190%), grpmax FAF 3.15e-06, 0 homozygotes
• gnomAD v4.1 exome: total AF 1.99861e-06 (0.00020%, 3/1501040 alleles), max NFE AF 2.72229e-06 (0.00027%), grpmax FAF 7.3e-07, 0 homozygotes
• gnomAD-Canada v1.0: absent (0 carriers)
• Threshold: generic PM2 (AF < 0.1% in population databases) per local non-VCEP gnomAD calibration; framework: Richards et al. 2015, PMID 25741868. Max observed AF 0.00190% < 0.1%.
• PM2 supporting-strength calibration: SVI Bayesian framework, Tavtigian et al. 2018, PMID 28492532.

Still missing

• gnomAD v2.1 and v4.1 report no genome (WGS) allele counts for this variant (genome AC=0, AN=0 in both releases); frequency is based on exome data only.
• PM2 applied at supporting strength only; no higher-strength population evidence available (e.g., large case-control absence data).

Sources

Reasoning

BP3 (Richards et al. 2015, PMID:25741868) applies to in-frame insertions/deletions in a repeat region without known function (supporting, benign direction). Conditions: (1) In-frame indel - met: c.585_587dup is an in-frame 3-nt duplication (p.Glu198dup, protein 384 -> 385 aa); no frameshift, no premature termination, so no NMD; SpliceAI max delta = 0.01 confirms no splice impact, so the variant acts purely at the protein level. (2) Repeat region - met: reference CDS analysis of NM_133509.4 shows the duplicated unit (c.585-587 = GGA) lies within a tandem trinucleotide repeat, (GAA)3 (codons 196-198 = GAA GAA GAA), encoding a homopeptide poly-glutamate low-complexity tract (EEE); the variant lengthens this repeated tract by one glutamic acid (E3 -> E4) - the classic single-amino-acid repeat-length change BP3 was designed for. (3) Without known function - met based on available evidence: residues 196-198 are not a characterized functional motif of RAD51B (the Walker A motif GPPGCGKT is at ~residues 109-116, outside this tract); no variant-specific functional data exist in the case (OncoKB: no variant-specific functional evidence, 'Unknown Oncogenic Effect'; literature_pass: 0 PMIDs; no publications in the case). BP3 is therefore met at supporting strength under the generic ACMG/AMP 2015 framework (no RAD51B VCEP/CSPEC exists; cspec found=false). Population note: gnomAD shows the allele is very rare (v2.1 2/219,854; v4.1 3/1,501,040; 0 homozygotes), so the common-polymorphic-repeat rationale does not apply, but BP3's structural conditions (in-frame indel in a repeat region without known function) are satisfied regardless. HUMAN REVIEW FLAG: the definition of 'repeat region' for a short 3-unit trinucleotide repeat (3-residue poly-Glu tract) is a threshold judgment; if a reviewer determines this tract is too short to qualify as a repeat region, BP3 would not apply and PM4 would instead become met (the two criteria are mutually exclusive for this variant).

Rule

acmg_generic_bp3: in_frame_indel_AND_repeat_region_AND_no_known_function

Evidence considered

• generic_acmg_combination_rules: governing generic ACMG/AMP 2015 criteria definitions (PMID:25741868); no RAD51B VCEP/CSPEC available so generic BP3 definition applies
• Reference CDS sequence analysis of the case transcript NM_133509.4 (codons 195-201: TTG GAA GAA GAA ATT ATC TCA): tandem trinucleotide repeat (GAA)3 encoding poly-Glu tract E3 at residues 196-198; c.585_587dupGGA (unit GGA, spanning the TTG|GAA boundary) expands the tract to E4, i.e., single-amino-acid repeat-length change in a low-complexity repeat region
• spliceai: max delta score 0.01 - no splice impact; the variant is a pure in-frame coding change (premise of BP3)
• pvs1_variant_assessment: consequence_class = in_frame_indel_unknown, variant_bucket = other, canonical_splice_consensus = false - confirms in-frame, non-null variant class
• gnomad_v2 / gnomad_v4: variant very rare with 0 homozygotes (v2.1 AF 9.10e-06, 2/219854; v4.1 AF 2.00e-06, 3/1501040) - confirms this is not a common polymorphic repeat-length allele; does not negate BP3 but removes the polymorphic-repeat rationale

Still missing

• No functional studies of the poly-Glu tract (residues 196-198) in RAD51B are available in the case evidence, so 'without known function' is supported by absence of functional annotation rather than direct negative evidence
• No VCEP/CSPEC specification exists for RAD51B; generic ACMG/AMP 2015 (PMID:25741868) BP3 definition used
• Repeat-region status of a (GAA)3 trinucleotide repeat / 3-residue homopeptide tract is a reviewer judgment (flagged needs_human_review=true)

Sources

Reasoning

PM4 (Richards et al. 2015, PMID:25741868) applies to protein length changes from in-frame deletions/insertions in a NON-repeat region, or stop-loss variants. NM_133509.4:c.585_587dupGGA is a 3-nt in-frame duplication in exon 7 (exon 7 spans c.573-756) producing NP_598193.2:p.(Glu198dup): the protein length changes from 384 to 385 amino acids (verified by translating the reference CDS of NM_133509.4 and applying the duplication; the poly-glutamate tract E3 (codons 196-198 = GAA GAA GAA) becomes E4). The protein-length-change condition is therefore met. However, the second condition fails: the change does NOT lie in a non-repeat region. The duplicated unit (CDS bases 585-587 = GGA) sits at the boundary of a tandem trinucleotide repeat, (GAA)3, encoding a homopeptide poly-glutamate low-complexity tract at residues 196-198; the net effect is lengthening of this repeated tract by one glutamic acid. PM4 is explicitly restricted to non-repeat regions, so it is not applied here; the repeat-region context routes this variant to BP3 instead (mutually exclusive determination). Stop-loss clause: not applicable (no stop-loss variant). No VCEP/CSPEC exists for RAD51B (cspec found=false, gene_vcep_dir=null), so the generic ACMG/AMP 2015 definitions apply (generic_acmg_combination_rules). Caveat: the classification of a 3-unit trinucleotide repeat as a 'repeat region' is a threshold judgment and is flagged for human review on BP3; if a reviewer rejects the repeat-region interpretation, PM4 would become met under the generic framework.

Rule

acmg_generic_pm4: in-frame_indel_protein_length_change_AND_non_repeat_region

Evidence considered

• generic_acmg_combination_rules: governing generic ACMG/AMP 2015 criteria definitions (PMID:25741868); no RAD51B VCEP/CSPEC available so generic PM4/BP3 definitions apply
• Reference CDS sequence analysis of the case transcript NM_133509.4 (codons 195-201: TTG GAA GAA GAA ATT ATC TCA; duplicated unit at c.585-587 = GGA): in-frame +1 aa, poly-Glu tract E3 (codons 196-198 = GAA GAA GAA) expands to E4 -> p.Glu198dup; consistent with Mutalyzer normalized protein (reference E-run 3 -> predicted E-run 4, length 384 -> 385 aa)
• pvs1_variant_assessment: consequence_class = in_frame_indel_unknown, variant_bucket = other, canonical_splice_consensus = false (no frameshift/stop-loss; no NMD trigger)
• spliceai: max delta score 0.01 - no splice impact, so the variant's sole consequence is the in-frame coding change evaluated here
• gnomad_v2 / gnomad_v4: variant is real but very rare (v2.1 AF 9.10e-06, 2/219854 alleles; v4.1 AF 2.00e-06, 3/1501040 alleles; 0 homozygotes) - population context only, not used to satisfy or refute PM4's structural conditions

Still missing

• No experimental/functional characterization of the poly-Glu tract (residues 196-198) or of this variant's effect on RAD51B function is available in the case evidence; 'without known function' and non-repeat/repeat-region classification rest on sequence context and absence of functional annotation.
• NM_133509.4 is not the MANE-select transcript (mane_select=false per VariantValidator annotation in prefetch.json); the in-frame consequence was verified on the reporting transcript and matches ClinVar's NM_133510.4 naming (p.Glu198_Ile199insGlu), but transcript-isoform-level context is otherwise unannotated in the case bundle.

Sources

Reasoning

Variant NM_133509.4:c.585_587dupGGA (RAD51B) is an in-frame duplication, p.(Glu198dup), in exon 7 (c.573-756) — not a missense SNV and not a synonymous variant, so the missense in-silico path (REVEL/BayesDel/aGVGD) does not apply and BP7 does not apply. Governing framework is generic ACMG/AMP 2015 (no CSPEC/VCEP found for RAD51B; vcep_materials.json: 'no cspec url available'). SpliceAI (the only applicable calibrated computational path) predicts no splice impact: max delta score = 0.01 (DS_AG 0.0, DS_AL 0.01, DS_DG 0.0, DS_DL 0.0), far below the 0.2 splice-impact threshold from the SpliceAI publication (Jaganathan et al., Cell 2019; PMID 30661751). No other calibrated computational evidence supports a deleterious effect: REVEL and BayesDel were skipped by the pipeline ('not snv or no coords') and BayesDel additionally has no verified published threshold available to this pipeline. The single ClinVar submission (VCV004149796, Ambry Genetics SCV006407454, 1-star, no validated PMIDs) asserts the variant 'is predicted to be deleterious by in silico analysis (choi y et al' (apparently PROVEAN, Choi et al. 2012; PMID 23066102) but provides no score and no threshold, so it is not usable as calibrated computational evidence under the named-publication threshold rule; this unverifiable claim is flagged for human review. PP3 therefore is not met.

Rule

Generic ACMG/AMP 2015 (PMID 25741868) PP3: multiple lines of computational evidence support a deleterious effect on the gene or gene product (conservation, evolutionary, splicing impact, etc.). Within this group, PP3 is evaluated via the calibrated computational-evidence paths — SpliceAI (splice-impact) and REVEL/BayesDel/aGVGD (missense) — using thresholds from named publications only.

Evidence considered

• SpliceAI max delta score = 0.01 (DS_AG 0.0, DS_AL 0.01, DS_DG 0.0, DS_DL 0.0) — below the 0.2 splice-impact threshold from the SpliceAI publication (Jaganathan et al., Cell 2019; PMID 30661751); no splice-impact support for PP3
• REVEL: no score available; tool skipped for non-SNV variant ('not snv or no coords'); REVEL missense calibration (Pejaver et al., Am J Hum Genet 2022; PMID 36413997) is not applicable to an in-frame duplication
• BayesDel: no score available; tool skipped ('not snv or no coords'); no verified published threshold is available to this pipeline, so any BayesDel value would be treated as insufficiently calibrated for PP3/BP4
• ClinVar VCV004149796 / SCV006407454 (Ambry Genetics): submitter-reported 'predicted to be deleterious by in silico analysis (choi y et al' — no score, no threshold, single 1-star submission with no validated PMIDs; not usable as calibrated computational evidence
• Variant class confirmed as in-frame duplication (p.Glu198dup; consequence_class 'in_frame_indel_unknown') via VariantValidator in prefetch.json and pvs1_variant_assessment.json, establishing that missense-only predictors and BP7 are inapplicable

Still missing

• ClinVar submission SCV006407454 (Ambry Genetics) asserts in-silico deleteriousness (PROVEAN, 'choi y et al') without providing a score or threshold; cannot be verified from available evidence — a human reviewer may need to query the submitter's actual PROVEAN score
• No calibrated in-silico predictor applicable to in-frame duplications is available in this pipeline (REVEL/BayesDel/aGVGD skipped: 'not snv or no coords'); only the splice-impact path (SpliceAI) was assessable

Sources

Reasoning

BA1 (allele frequency >1% in a reference population) is not met. The highest observed population allele frequency for NM_133509.4:c.585_587dupGGA is 1.89836e-05 (0.00190%) in European (non-Finnish) gnomAD v2.1 (2/105,354 alleles); gnomAD v4.1 total AF is 1.99861e-06 (0.00020%, 3/1,501,040 alleles) with NFE max 2.72229e-06 (0.00027%). These values are three orders of magnitude below the generic ACMG/AMP BA1 threshold of >1%. grpmax FAF (3.15e-06 v2.1; 7.3e-07 v4.1) is likewise far below threshold. gnomAD-Canada v1.0: absent. No disease-prevalence or penetrance argument could place this variant near BA1; generic ACMG/AMP fallback applies because no ClinGen CSPEC/VCEP or local gene-specific framework exists for RAD51B (framework_mode=generic_acmg).

Rule

Generic ACMG/AMP 2015 (PMID 25741868) BA1: allele frequency >1% in a large reference population is stand-alone benign. Not met at max observed AF 0.00190%.

Evidence considered

• gnomAD v2.1 exome: total AF 9.09695e-06 (0.00091%, 2/219854 alleles), max subpopulation NFE AF 1.89836e-05 (0.00190%), grpmax FAF 3.15e-06, 0 homozygotes
• gnomAD v4.1 exome: total AF 1.99861e-06 (0.00020%, 3/1501040 alleles), max subpopulation NFE AF 2.72229e-06 (0.00027%), grpmax FAF 7.3e-07, 0 homozygotes
• gnomAD-Canada v1.0: absent
• Threshold: generic ACMG/AMP 2015 BA1 (AF > 1% in population databases) per local generic calibration for non-VCEP gnomAD use; framework: Richards et al. 2015, PMID 25741868. Max observed AF 0.00190% does not approach 1%.

Sources

Reasoning

BS1 (allele frequency greater than expected for the disorder) is not met. The highest population allele frequency observed is 0.00190% (NFE, gnomAD v2.1), and gnomAD v4.1 reports only 0.00027% in the same subpopulation. Both are far below the generic BS1 threshold of >0.3% used for non-VCEP gnomAD evaluation. Even under the most permissive modeling of RAD51B-associated cancer susceptibility (dominant, incomplete penetrance, adult onset; no prevalence estimate available in this case), an AF of 0.0019% cannot exceed the expected disease-allele frequency. No BA1/BS1 threshold artifact is present: the variant is a 3-bp in-frame duplication reliably genotyped in exomes, with consistent allele counts across two independent gnomAD releases.

Rule

Generic ACMG/AMP 2015 (PMID 25741868) BS1: allele frequency greater than expected for the disorder. Not met at max AF 0.00190% (<0.3% local threshold).

Evidence considered

• gnomAD v2.1 exome: max subpopulation AF 1.89836e-05 (0.00190%, NFE, 2/105354 alleles)
• gnomAD v4.1 exome: max subpopulation AF 2.72229e-06 (0.00027%, NFE, 3/1102012 alleles)
• gnomAD-Canada v1.0: absent
• Threshold: generic BS1 (AF greater than expected for the disorder; local non-VCEP calibration >0.3%) per Richards et al. 2015, PMID 25741868. Max observed AF 0.00190% is below 0.3%.

Sources

Reasoning

BS2 (observed in a healthy adult) is not met. Zero homozygotes are observed for this variant in gnomAD v2.1 (0/219,854 alleles) and gnomAD v4.1 (0/1,501,040 alleles); for a recessive disease, BS2 would require homozygous healthy adults, and RAD51B-related cancer susceptibility is modeled as a dominant adult-onset trait with incomplete penetrance, for which BS2 requires full penetrance at an early age — not the case here. The 5 heterozygous carriers in gnomAD are population controls and cannot serve as BS2 evidence for an incompletely penetrant, late-onset cancer-predisposition phenotype. No curated unaffected-carrier or case-control clinical data are present in this case (literature_pass total_pmids=0; no ClinVar population/carrier data; ClinVar entry 4149796 is a single-submitter VUS with no frequency information).

Rule

Generic ACMG/AMP 2015 (PMID 25741868) BS2: observed in a healthy adult (homozygous for recessive, or heterozygous with full penetrance expected at an early age for dominant). Not met: 0 homozygotes and no full-penetrance early-onset dominant phenotype.

Evidence considered

• gnomAD v2.1 exome: 0 homozygotes (2/219854 alleles)
• gnomAD v4.1 exome: 0 homozygotes (3/1501040 alleles)
• ClinVar variation 4149796: single 1-star germline submission, Uncertain significance (Ambry Genetics, SCV006407454); no frequency, homozygote, or healthy-carrier data; not usable as BS2 evidence
• RAD51B germline context: dominant adult-onset cancer susceptibility with incomplete penetrance (per pvs1_gene_context disease context), so the BS2 dominant-disease clause (full penetrance at early age) does not apply

Still missing

• No curated clinical unaffected-carrier or case-control dataset available in this case to directly assess BS2 (literature_pass found 0 PMIDs).

Sources

Reasoning

BP4 requires multiple lines of computational evidence suggesting no impact on the gene or gene product. The only available computational line is SpliceAI, which predicts no splice impact (max delta score = 0.01, below the 0.2 threshold from Jaganathan et al., Cell 2019; PMID 30661751). This single line addresses splice impact only and does not establish that the variant has no impact on the gene product: the variant is an in-frame duplication (p.Glu198dup) that itself alters the protein sequence, and no calibrated predictor for in-frame indel protein-level impact exists in this pipeline (REVEL/BayesDel/aGVGD were skipped: 'not snv or no coords'). The 'multiple lines' bar of BP4 is therefore not satisfied; BP4 is not met.

Rule

Generic ACMG/AMP 2015 (PMID 25741868) BP4: multiple lines of computational evidence suggest no impact on the gene or gene product. Within this group, BP4 is evaluated via the same calibrated computational-evidence paths as PP3 (SpliceAI for splice impact; REVEL/BayesDel/aGVGD for missense), using thresholds from named publications only.

Evidence considered

• SpliceAI max delta score = 0.01 (DS_AG 0.0, DS_AL 0.01, DS_DG 0.0, DS_DL 0.0) — below the 0.2 splice-impact threshold from the SpliceAI publication (Jaganathan et al., Cell 2019; PMID 30661751): no splice impact predicted, but this is a single computational line covering splice impact only
• REVEL/BayesDel/aGVGD: not applicable and no scores available — variant is an in-frame duplication, not a missense SNV ('not snv or no coords'); no calibrated predictor for this variant class
• Variant class: in-frame duplication (p.Glu198dup; consequence_class 'in_frame_indel_unknown') per prefetch.json (VariantValidator) and pvs1_variant_assessment.json — the variant itself alters the protein, so a negative splice prediction cannot by itself establish 'no impact on gene or gene product'

Still missing

• Only one computational line is available (SpliceAI, negative for splice impact); BP4 requires multiple lines suggesting no impact on gene or gene product
• SpliceAI addresses splice impact only; the in-frame duplication itself alters the protein (p.Glu198dup), and no calibrated predictor for in-frame indel protein-level impact is available (REVEL/BayesDel/aGVGD skipped: 'not snv or no coords')

Sources

Reasoning

No proband-level de novo evidence exists for this case. The evidence bundle contains no proband phenotype/pedigree record, no parental testing results, and no documented de novo occurrence (confirmed or assumed) of NM_133509.4:c.585_587dup (p.Glu198dup). The only variant-specific clinical record, ClinVar VCV004149796 (single submitter, Ambry Genetics, SCV006407454, 'criteria provided, single submitter'), carries no de novo or family information and no PMIDs. No ClinGen CSPEC/VCEP specification exists for RAD51B, so generic ACMG/AMP 2015 (PMID 25741868) governs. PS2 cannot be applied in the absence of any de novo observation.

Rule

PS2 (Richards et al. 2015, PMID 25741868): de novo variant (paternity and maternity confirmed) - strong evidence of pathogenicity.

Evidence considered

• clinvar

Still missing

• No proband de novo observation data (trio or parental testing) in the case bundle
• No ClinVar submission or literature report documents a confirmed de novo occurrence of c.585_587dup
• No paternity/maternity confirmation data available

Sources

Reasoning

PS3 requires well-established functional studies demonstrating a damaging effect on the gene or gene product (generic ACMG/AMP 2015, PMID 25741868; no RAD51B CSPEC/VCEP exists for this case). No functional evidence of any kind is available for NM_133509.4:c.585_587dup (p.(Glu198dup)): the literature pass returned 0 PMIDs, the publications directory is empty, OncoKB explicitly states there is no variant-specific reviewed functional evidence (classification 'Unknown Oncogenic Effect'), COSMIC has no somatic report of the variant, and the single ClinVar submission (VCV004149796, Ambry Genetics, VUS) contains no functional assay data - only an in silico PP3 signal. SpliceAI predicts no splice impact (max delta score 0.01), but this is an in silico prediction, not a functional study, and cannot substitute for PS3 evidence. With no assay data testing this exact variant, a damaging functional effect cannot be established; the criterion is therefore not_assessed pending functional evidence.

Rule

PS3 (generic ACMG/AMP 2015, PMID 25741868): Well-established functional studies show a damaging effect on the gene or gene product.

Evidence considered

• generic_acmg_combination_rules - Richards et al. 2015 ACMG/AMP sequence variant interpretation framework (PMID 25741868): PS3 requires well-established functional studies demonstrating a damaging effect on the gene or gene product; no such studies exist for this variant
• oncokb - OncoKB variant page for RAD51B E198dup explicitly states no variant-specific reviewed functional evidence was identified; classification 'Unknown Oncogenic Effect'
• clinvar - VCV004149796 (SCV006407454, Ambry Genetics, single submitter, Uncertain significance) contains no functional assay data; only an in silico (PP3) signal
• spliceai - max delta score 0.01 (no significant splice impact predicted); in silico prediction only, not a functional study, so it cannot meet PS3 evidentiary standards

Still missing

• No well-established functional assay data for p.(Glu198dup) - e.g., homologous recombination (HR) repair assays, RAD51B paralog interaction/function assays, or cell-based DNA-damage survival assays testing this exact variant
• No variant-specific functional publications identified (literature_pass total_pmids = 0; fulltext_map.json empty)
• No validated RAD51B-specific functional assay with calibrated PS3/BS3 thresholds available in this case

Sources

Reasoning

PS4 (Richards et al. 2015, PMID:25741868) requires the prevalence of the variant in affected individuals to be statistically significantly increased over controls (case-control enrichment). No case-control study of NM_133509.4:c.585_587dup (NP_598193.2:p.(Glu198dup)) in RAD51B exists in this case's evidence bundle: literature_pass.json reports total_pmids = 0 with the note 'No PMIDs found across evidence sources.', fulltext_map.json is empty, and no publications directory was retrieved. gnomAD v2.1 (AF 9.1e-06, 2/219854 alleles) and v4.1 (AF 2.0e-06, 3/1501040 alleles) provide population frequency only and cannot substitute for a case-control comparison. Evidence is insufficient to evaluate PS4, so it is returned as not_assessed rather than met/not_met.

Evidence considered

• literature_pass.json: total_pmids = 0, criteria_index = {}, note 'No PMIDs found across evidence sources.'
• gnomAD v2.1 total AF = 9.09695e-06 (2/219854 alleles, 0 homozygotes), grpmax FAF = 3.15e-06 (evidence.json GNOMAD_V2_1) - population frequency only, not case-control data
• gnomAD v4.1 total AF = 1.99861e-06 (3/1501040 alleles, 0 homozygotes), grpmax FAF = 7.3e-07 (evidence.json GNOMAD_V4_1) - population frequency only, not case-control data
• ClinVar VCV004149796 submission audit: 1 submission, Uncertain significance, 0 validated PMIDs, 0 criterion leads (prefetch.json clinvar.submission_audit)

Still missing

• No case-control study of this exact variant (NM_133509.4:c.585_587dup / p.Glu198dup) is available; literature_pass identified 0 PMIDs and no publications directory exists for the case. A variant-level case-control enrichment study would be required to evaluate PS4.

Sources

Reasoning

No governing VCEP/CSPEC exists for RAD51B (cspec found: false; framework_mode: generic_acmg), so the generic ACMG/AMP 2015 rule applies (PMID 25741868): PM3 requires, for a recessive disorder, that the variant be detected in trans with a pathogenic variant. The consolidated evidence bundle contains no case-level data that could establish this: (1) pvs1_gene_context records inheritance as [] and no recessive-disease context for RAD51B is established in any retrieved framework material; (2) ClinVar VCV004149796 contains a single submission (Ambry Genetics) classified Uncertain significance with no pathogenic variant, no phase, and no trans/cis configuration reported (usable_for_criterion_leads: false, no PMIDs); (3) literature_pass.json found zero PMIDs and fulltext_map is empty, so no publication reports this variant in trans with a pathogenic allele; (4) gnomAD v2.1/v4.1 frequencies (AF 9.1e-06 and 2.0e-06, 0 homozygotes) are population observations only and carry no allelic configuration information. Because no evidence of an in-trans pathogenic variant (nor any genotype/phase data) exists, PM3 cannot be met or refuted from the current bundle; it is returned not_assessed pending proband genotyping, phase determination, and confirmation of the relevant inheritance mode.

Rule

PM3 (ACMG/AMP 2015, PMID 25741868): For recessive disorders, detected in trans with a pathogenic variant.

Evidence considered

• generic_acmg_combination_rules (ACMG/AMP 2015, PMID 25741868): PM3 applies only for recessive disorders where the variant is observed in trans with a pathogenic variant; no such observation exists in this case
• clinvar (VCV004149796, ClinVar variation 4149796): single submitter (Ambry Genetics), Uncertain significance, germline, criteria provided but not usable for criterion leads (no valid PMIDs); no pathogenic variant, phase, or trans/cis data reported
• pvs1_gene_context: germline disease context present but inheritance list empty; no recessive phenotype/RAD51B specification establishes a recessive mode relevant to PM3
• gnomad_v2 / gnomad_v4: population allele frequencies (2/219854 and 3/1501040 alleles, 0 homozygotes) provide no allelic/phase information

Still missing

• No proband/pedigree genotyping data: no second RAD51B variant reported in any affected individual, so trans configuration with a pathogenic/likely pathogenic variant cannot be evaluated
• No phase determination (phasing with a second RAD51B alteration) available in any source in the bundle
• Inheritance mode of the RAD51B-associated phenotype not established in governing materials (pvs1_gene_context inheritance: []); recessive-disease context required by generic PM3 is unconfirmed
• No variant-specific literature: literature_pass total_pmids = 0 and fulltext_map empty; no publication reports this exact variant in trans with a pathogenic variant

Sources

Reasoning

No de novo observation is available at any confidence level. PM6 requires an assumed de novo occurrence (without confirmed maternity/paternity); the case bundle contains no proband-parent trio data, no segregation of the variant to a proband from unaffected parents, and no literature report of c.585_587dup arising de novo (literature_pass total_pmids=0; fulltext_map empty). The single ClinVar submission (VCV004149796) contains no de novo assertion. PM6 cannot be applied.

Rule

PM6 (Richards et al. 2015, PMID 25741868): assumed de novo, without confirmation of paternity and maternity - moderate evidence of pathogenicity.

Evidence considered

• clinvar

Still missing

• No assumed-de-novo observation data (e.g., proband heterozygous with reportedly unaffected parents, parental relationships unconfirmed)
• No literature or ClinVar report of de novo occurrence for this variant

Sources

Reasoning

No segregation data are available. No affected relatives, informative meioses, or cosegregation counts for c.585_587dup are reported anywhere in the case bundle (case_summary, prefetch.json, evidence.json, literature_pass.json contain no pedigree or family-genotype information). The ClinVar record (VCV004149796) provides no segregation evidence and no PMIDs. Gene-level RAD51B germline papers collected for PVS1 context (e.g., PMID:25600502) do not mention this exact variant and are therefore not usable under variant-grounded citation rules. PP1 cannot be applied.

Rule

PP1 (Richards et al. 2015, PMID 25741868): co-segregation with disease in affected family members - supporting evidence of pathogenicity.

Evidence considered

• clinvar

Still missing

• No family segregation data (affected relatives genotyped for the variant)
• No meioses/cosegregation counts available for any family
• No literature report of this exact variant segregating with disease

Sources

Reasoning

PP4 (Richards et al. 2015, PMID:25741868) requires a phenotype or family history highly specific for a gene with a single genetic etiology. The case bundle contains no proband phenotype, family history, or segregation data: review_manifest.json and case_summary.json carry no clinical phenotype fields, and searches for phenotype/proband/segregation terms across the bundle returned nothing. In addition, the RAD51B germline disease context retrieved in pvs1_gene_context.json is cancer predisposition (breast/ovarian cancer susceptibility, cutaneous melanoma), which is not a highly specific single-etiology phenotype, so PP4 would rarely be supportable even with phenotype data. Evidence is insufficient to evaluate PP4; returned as not_assessed.

Evidence considered

• review_manifest.json / case_summary.json: no proband phenotype, family history, or segregation fields present
• pvs1_gene_context.json: RAD51B germline disease context is cancer predisposition (disease_names: 'Germline RAD51B truncating mutation in a family with cutaneous melanoma.'), not a highly specific single-etiology phenotype

Still missing

• No proband phenotype or family history information is provided anywhere in the case bundle; clinical phenotype data would be required to evaluate PP4.

Sources

Reasoning

BS3 requires well-established functional studies showing no damaging effect on protein function or splicing (generic ACMG/AMP 2015, PMID 25741868; no RAD51B CSPEC/VCEP exists for this case). No functional assay data exist for p.(Glu198dup): the literature pass returned 0 PMIDs, OncoKB reports no variant-specific reviewed functional evidence, the ClinVar submission carries no functional data, and COSMIC has no somatic report. SpliceAI predicts no significant splice impact (max delta 0.01), but in silico predictions do not qualify as well-established functional studies and cannot satisfy BS3. The very low population allele frequency (gnomAD v2.1 AF 9.10e-06; v4.1 AF 2.00e-06) is population-frequency evidence adjudicated under BS1/BA1 by the population group and does not substitute for BS3. With no assay demonstrating retained/normal function for this exact variant, BS3 cannot be established; the criterion is not_assessed pending functional evidence.

Rule

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

Evidence considered

• generic_acmg_combination_rules - Richards et al. 2015 ACMG/AMP sequence variant interpretation framework (PMID 25741868): BS3 requires well-established functional studies showing no damaging effect on protein function or splicing; no such studies exist for this variant
• oncokb - OncoKB variant page for RAD51B E198dup identifies no variant-specific reviewed functional evidence; nothing demonstrating normal function
• spliceai - no significant splice impact predicted (max delta 0.01), but in silico predictions do not qualify as well-established functional studies for BS3
• clinvar - single-submitter Uncertain significance submission with no functional evidence; cannot support BS3

Still missing

• No functional assay (e.g., homologous recombination repair or RAD51B paralog function assay) demonstrating normal function for p.(Glu198dup)
• No variant-specific functional publications identified (literature_pass total_pmids = 0)
• No RAD51B CSPEC/VCEP or gene-specific PS3/BS3 calibration to reference

Sources

Reasoning

No non-segregation data are available. BS4 requires demonstration that affected family members do not carry the variant; no family genotyping or segregation data exist in the case bundle, and the ClinVar record (VCV004149796) contains none. Low gnomAD allele counts (v2.1: 2/219854, AF 9.10e-06; v4.1: 3/1501040, AF 2.00e-06) are population-frequency observations relevant to population-frequency criteria (e.g., BA1/BS1/PM2, assessed by other groups), not to BS4 non-segregation within affected families, and are not used here. BS4 cannot be applied.

Rule

BS4 (Richards et al. 2015, PMID 25741868): non-segregation with disease - strong evidence of benign impact.

Evidence considered

• clinvar

Still missing

• No data showing an affected family member lacking the variant (non-segregation)
• No family/pedigree genotyping results in the case bundle or ClinVar

Sources

Reasoning

Generic ACMG/AMP 2015 rule applies (no RAD51B VCEP/CSPEC): BP2 requires observing the variant in trans with a pathogenic variant for a fully dominant disease/gene, or in cis with a pathogenic variant in any inheritance pattern. The bundle contains no data supporting either configuration: the single ClinVar submission (VCV004149796, Ambry Genetics, Uncertain significance) reports no second RAD51B variant, no phase, and no cis/trans placement; no proband or family genotype data are present anywhere in the bundle; literature_pass.json returned zero PMIDs and fulltext_map is empty, so no publication reports this exact variant in cis or in trans with a pathogenic variant. gnomAD frequencies (v2.1 AF 9.1e-06; v4.1 AF 2.0e-06; 0 homozygotes) are unphased population observations. Without genotype/phase data from an observed individual, BP2 can be neither met nor refuted; it is returned not_assessed pending case-level phasing information.

Rule

BP2 (ACMG/AMP 2015, PMID 25741868): Observed in trans with a pathogenic variant for a fully dominant disease/gene; or observed in cis with a pathogenic variant in any inheritance pattern.

Evidence considered

• generic_acmg_combination_rules (ACMG/AMP 2015, PMID 25741868): BP2 requires observation in trans with a pathogenic variant for a fully dominant disease/gene, or in cis with a pathogenic variant; no such observation is present
• clinvar (VCV004149796, ClinVar variation 4149796): single submission (Ambry Genetics) Uncertain significance; no pathogenic variant, no phase, no cis/trans configuration reported; no PMIDs
• gnomad_v2 / gnomad_v4: unphased population allele counts (2/219854 and 3/1501040 alleles, 0 homozygotes); no allelic-configuration evidence

Still missing

• No proband/family genotype data in the bundle; the variant has not been observed in cis or in trans with any pathogenic variant
• No phase determination available from ClinVar, literature, or any other source in the bundle
• No variant-specific literature: zero PMIDs across evidence sources (literature_pass total_pmids = 0, fulltext_map empty)
• Dominant disease/gene context for RAD51B is not formally specified in governing materials, and no observed trans-pathogenic configuration exists to test BP2

Sources

Reasoning

BP5 (Richards et al. 2015, PMID:25741868) applies when a variant is found in a case with an alternate molecular basis of disease (e.g., a documented pathogenic variant in another gene that explains the phenotype). The case bundle contains no proband-level genetic testing results, no co-occurrence data, and no documented alternate molecular diagnosis: review_manifest.json and case_summary.json contain no clinical or case-level fields, and literature_pass.json identified 0 PMIDs. Evidence is insufficient to evaluate BP5; returned as not_assessed.

Evidence considered

• review_manifest.json / case_summary.json: no proband clinical data, co-occurrence data, or alternate-diagnosis fields present
• literature_pass.json: total_pmids = 0 (no source reports co-occurrence or alternate-molecular-basis data for this variant)

Still missing

• No proband co-occurrence or alternate-molecular-basis data are available; BP5 requires evidence of a separate pathogenic molecular diagnosis explaining the phenotype in the same case.

Sources