CDKN2A c.170C>G, p.(Ala57Gly)

Reasoning

Generic ACMG/AMP 2015 (Richards et al., PMID:25741868) governs (no CDKN2A CSPEC/VCEP or local custom framework; generic_acmg_combination_rules is the governing PP3/BP4 calibration for this case). This is a missense variant (NM_000077.5:c.170C>G, p.(Ala57Gly)), so exactly one computational mechanism is evaluated - the REVEL missense path per the calibration document. REVEL = 0.205 (revel registry: local REVEL v1.3 lookup), which is below the governing BP4 threshold of <0.250 defined in generic_acmg_combination_rules, so the criterion is met at supporting strength. The prediction is counted once (BP4 only); it is not double-counted toward PP3 or BP7. BayesDel (-0.234048) is not used (no verified published threshold per the governing calibration). The SpliceAI score (max delta 0.00) is not applied to this missense variant per the one-mechanism-per-variant rule, so it neither adds nor conflicts with the BP4 call. ClinVar single-submitter statements describing in-silico tolerance (GeneDx SCV000568018, Ambry SCV000217810, Color SCV000903860) are directionally consistent with a neutral computational prediction but were not used as independent evidence. No literature source provides a gene-specific computational assessment of p.A57G (PMID:21462282 lists it only as a VUS without computational scores; PMID:19260062 in-silico results concern the same-codon p.Ala57Val only).

Rule

ACMG/AMP 2015 BP4 (PMID:25741868): multiple lines of computational evidence suggest no impact on gene or gene product. Governing generic-fallback calibration (generic_acmg_combination_rules): for missense variants, REVEL < 0.250 -> BP4 (supporting). Met at supporting strength: REVEL = 0.205.

Evidence considered

• Governing framework: generic ACMG/AMP 2015 fallback (Richards et al., PMID:25741868) because no ClinGen CSPEC/VCEP or local custom CDKN2A specification exists in this case (cspec found=false; vcep_materials.json empty; gene_vcep_dir=null).
• Variant class: missense substitution NM_000077.5:c.170C>G, NP_000068.1:p.(Ala57Gly) (exon 2). Per the governing calibration (generic_acmg_combination_rules, output/generic_acmg_classification_rules.md, 'PP3/BP4 in-silico calibration (generic, non-VCEP only)'): missense variants use REVEL only; REVEL < 0.250 -> BP4 (supporting); REVEL in [0.250, 0.750] -> neither met.
• REVEL score 0.205 (source 'revel': local REVEL v1.3 lookup, chrom 9 pos 21971189 G>C). 0.205 < 0.250, therefore BP4 is met at supporting strength.
• The REVEL prediction is used exactly once - for BP4 - and is not double-counted across PP3/BP4/BP7, consistent with the calibration document's instruction never to apply both mechanisms independently or count the same prediction more than once.
• BayesDel score -0.234048 (source 'bayesdel') is not used: the governing calibration states no verified published BayesDel threshold is available to this pipeline, so the score is treated as insufficiently calibrated regardless of any recalled cutoff.
• SpliceAI max delta score = 0.00 (source 'spliceai') is not applied to this missense variant per the one-mechanism-per-variant rule ('a missense variant's own SpliceAI score does not clear its REVEL-assessed protein-level effect'); therefore it is not used to support BP4 either, avoiding any risk of double-counting distinct predictor paths for one variant.
• Literature check: PMID:21462282 lists p.A57G in Table 1C (variants of uncertain significance, 1 MPM carrier, 0 FM families) without any computational score for the variant; PMID:19260062 reports in-silico neutral predictions only for the same-codon p.Ala57Val (c.170C>T). Neither provides a gene-specific pre-assigned code for p.Ala57Gly; no extrapolation was made.
• ClinVar single-submitter in-silico comments (GeneDx SCV000568018, Ambry SCV000217810 'predicted to be tolerated', Color SCV000903860 'may not impact protein structure and function') were reviewed for consistency but were not used as evidence, which keeps the BP4 call grounded in the directly retrieved REVEL score and the governing calibration.

Still missing

• BP4 rests on a single calibrated missense predictor (REVEL 0.205 < 0.250) as defined by the generic-fallback calibration; there is no CDKN2A-specific VCEP/CSPEC rule to corroborate or refine the threshold for this gene.
• BayesDel -0.234048 could not be used to corroborate BP4 because the governing calibration identifies no verified published BayesDel threshold.

Sources

Reasoning

No case-control or cohort study demonstrates a significantly increased prevalence of the exact variant CDKN2A c.170C>G (p.Ala57Gly) in affected individuals. Available variant-level observations point away from enrichment: in the Norwegian population-based multiple primary melanoma (MPM) case-control material (PMID:22447455, which re-describes the CDKN2A screening of the same cohort first published as PMID:18023021), Ala57Gly was detected in one subject of the 420-person healthy blood-donor control population and is not reported among the CDKN2A mutations carried by the 388 MPM patients (27 patients carried 10 different CDKN2A mutations). The CDKN2A clinico-epidemiologic variant database (PMID:21462282) lists p.A57G as a VUS with a single MPM carrier and 0 familial-melanoma families, 0 sporadic melanoma and 0 nonmelanoma carriers - an anecdotal observation without segregation, far below PS4-grade statistical evidence. ClinVar (variation 187272) contains only single-submitter clinical laboratory Uncertain significance submissions (7 labs), which are not case-control evidence and cannot establish PS4. The variant is additionally present in the general population (gnomAD v4.1: 22/1,597,598 alleles, 0 homozygotes; grpmax FAF ~1.1e-5). PS4 is therefore not met under the generic ACMG/AMP 2015 framework (no CDKN2A CSPEC/VCEP available).

Rule

PS4 (Richards et al. 2015): prevalence of the variant in affected individuals significantly increased compared with prevalence in controls; not met when only anecdotal case observations exist and the variant is also observed in unaffected controls.

Evidence considered

• Generic ACMG/AMP 2015 fallback governs this case (Richards et al. 2015, PMID 25741868); no CDKN2A CSPEC/VCEP was retrieved (cspec found=false, framework_mode=generic_acmg).
• PMID:22447455 (Helsing et al. 2012): population-based MPM case-control material (388 Norwegian MPM patients vs 420 anonymous blood-donor controls). Direct quote: 'One CDKN2A mutation (Ala57Gly) was detected in one subject of the control population.' The 27 CDKN2A-mutation carriers among patients carried 10 different mutations, and Ala57Gly is not stated to be among them - the variant therefore shows no case enrichment.
• PMID:21462282 (Miller et al. 2011): CDKN2A Mutation Database clinical-epidemiologic listing (Table 1C) counts p.A57G as 0 familial-melanoma families, 1 MPM carrier, 0 sporadic melanoma carriers, 0 nonmelanoma carriers, 1 total carrier, i.e., a VUS with a single anecdotal affected observation and no segregation.
• ClinVar variation 187272: aggregate classification Uncertain significance, 8 submissions all 'criteria provided, single submitter' clinical laboratories (7 usable), 0 expert-panel submissions; laboratory VUS observations were not treated as case-control enrichment evidence for PS4.
• gnomAD v4.1 shows the variant in 22/1,597,598 alleles (0 homozygotes, grpmax FAF 1.108e-5), consistent with absence of strong case enrichment, noted as context only (population-frequency criteria are adjudicated by their own group).

Still missing

• No formal case-control or cohort study with statistical testing (OR/enrichment) for the exact variant c.170C>G was identified; only anecdotal case observations (1 MPM carrier) and a single control observation (1/420) exist.
• ClinVar submissions are all Uncertain significance from single-submitter laboratories; no pathogenic assertion or expert-panel review of this variant exists to leverage for prevalence.

Sources

Reasoning

Generic ACMG/AMP 2015 (Richards et al., PMID:25741868) governs because no ClinGen CSPEC/VCEP or local custom gene framework exists for CDKN2A (vcep_materials.json empty; cspec found=false; gene_vcep_dir=null; final_classification_framework found=false). The governing PP3/BP4 in-silico calibration for the generic fallback (generic_acmg_combination_rules: output/generic_acmg_classification_rules.md) specifies exactly one mechanism per variant matched to variant type: this variant is a missense substitution (NM_000077.5:c.170C>G, NP_000068.1:p.(Ala57Gly)), so only the REVEL missense path is evaluated. REVEL = 0.205 (revel registry: local REVEL v1.3 lookup), which does NOT exceed the PP3 threshold of >0.750 defined in the governing calibration document, so no computational evidence supports a deleterious effect. BayesDel score (-0.234048) is treated as not usable for PP3/BP4 because the governing calibration states no verified published threshold is available to this pipeline (BayesDel-specific rule), and recalled cutoffs without a citable publication cannot be applied. The SpliceAI path is not applied to this missense variant per the calibration document ('a missense variant's own SpliceAI score does not clear its REVEL-assessed protein-level effect'; splice path reserved for intronic/synonymous/non-canonical-splice-position variants); for context only, SpliceAI max delta = 0.00, far below any splice-impact threshold, so no PP3 splice contribution exists either. Literature relevant to gene-specific computational pre-assignment was checked: PMID:21462282 (computational + laboratory CDKN2A classification study) reports no computational score for p.A57G (Table 1C VUS listing only), and PMID:19260062 reports in-silico results only for the same-codon different variant p.Ala57Val (c.170C>T), which is not extrapolated to p.Ala57Gly. PP3 is not met.

Rule

ACMG/AMP 2015 PP3 (PMID:25741868): multiple lines of computational evidence support a deleterious effect on the gene or gene product. Governing generic-fallback calibration (generic_acmg_combination_rules): for missense variants, REVEL > 0.750 -> PP3 supporting. Not met: REVEL = 0.205.

Evidence considered

• Governing framework: generic ACMG/AMP 2015 fallback (Richards et al., PMID:25741868) because vcep_materials.json, prefetch cspec lookup, and final_classification_framework.json found no ClinGen CSPEC/VCEP or local custom CDKN2A specification (cspec found=false; gene_vcep_dir=null).
• Variant class: missense substitution NM_000077.5:c.170C>G, NP_000068.1:p.(Ala57Gly) (exon 2). Under the governing PP3/BP4 calibration (generic_acmg_combination_rules, 'PP3/BP4 in-silico calibration (generic, non-VCEP only)'), exactly one mechanism is evaluated per variant matching the variant's own type; missense variants are assessed with REVEL only, and the SpliceAI path is reserved for intronic/synonymous/non-canonical-splice-position variants.
• REVEL score 0.205 (source 'revel': local REVEL v1.3 lookup, chrom 9 pos 21971189 G>C). The generic_acmg_combination_rules calibration (output/generic_acmg_classification_rules.md) defines REVEL > 0.750 -> PP3 (supporting) for missense variants; 0.205 does not exceed 0.750, so PP3 is not met.
• BayesDel score -0.234048 (source 'bayesdel': local noAF chr9 lookup) was retrieved but is NOT usable for PP3/BP4: the governing calibration document states 'BayesDel: no verified published threshold is available to this pipeline. Never used for PP3/BP4 under the generic fallback' - any recalled cutoff lacking a named publication would be indistinguishable from a fabricated number and is therefore treated as not_available.
• SpliceAI (source 'spliceai': SpliceAI Lookup, NM_000077.5:c.170C>G) max delta score = 0.00. Not applied to this missense variant under the generic_acmg_combination_rules one-mechanism rule; recorded for context only (it provides no splice-impact support for PP3 in any case).
• Literature check for gene-specific computational pre-assignment: PMID:21462282 (Classifying variants of CDKN2A using computational and laboratory studies) was reviewed - p.A57G appears only in Table 1C as a clinico-epidemiologic VUS with no computational score reported for it; PMID:19260062 reports in-silico predictions only for the same-codon different substitution p.Ala57Val (c.170C>T), which cannot be extrapolated to p.Ala57Gly. No source pre-assigns a deleterious computational code for this variant.
• ClinVar single-submitter in-silico comments (e.g., SCV000568018 GeneDx 'in silico analysis supports that this missense variant does not alter protein structure/function'; SCV000903860 Color 'computational prediction tool suggests that this variant may not impact protein structure and function') are laboratory statements, not primary predictor evidence, and were not used to call PP3 or BP4.

Still missing

• REVEL 0.205 is far below the governing PP3 threshold (>0.750, generic_acmg_combination_rules); no calibrated computational predictor supports a deleterious effect for this variant.
• BayesDel -0.234048 could not be considered because the governing calibration document identifies no verified published BayesDel threshold; a citable calibration would be required before any BayesDel score could inform PP3.
• No CDKN2A-specific VCEP/CSPEC PP3 rule exists to pre-assign a computational code for this variant; only the generic fallback calibration was available.

Sources

Reasoning

Under operating policy, PP5 is triggered only by an exact-variant ClinVar expert-panel classification of Pathogenic or Likely pathogenic. The ClinVar submission audit for NM_000077.5(CDKN2A):c.170C>G (p.Ala57Gly) (variation 187272) shows expert_panel_submissions = 0: all 8 submissions are 'criteria provided, single submitter' clinical laboratory entries, and the aggregate classification is Uncertain significance (7 clinical laboratories; review status 'criteria provided, multiple submitters, no conflicts', 2 stars). No expert-panel pathogenic or likely-pathogenic assertion exists for this variant, and no reputable-source pathogenic assertion is available to the laboratory; PP5 must never be triggered by ordinary laboratory submissions or aggregate ClinVar labels. PP5 is therefore not met.

Rule

PP5 (Richards et al. 2015), restricted by operating policy: reputable-source (ClinVar expert-panel) recent report of the exact variant as Pathogenic/Likely pathogenic, without independent laboratory evidence.

Evidence considered

• Operating policy for this pipeline: PP5 applies only for an exact-variant ClinVar expert-panel classification of Pathogenic/Likely pathogenic; ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions never trigger PP5.
• ClinVar submission audit (variation 187272, VCV000187272): submissions_total=8, expert_panel_submissions=0, usable_for_classification=7; every submission is a single-submitter clinical laboratory (GeneDx, Sema4, Baylor Genetics, Labcorp, Ambry Genetics, Skane University Hospital Lund, Color Health, Quest Diagnostics) classified Uncertain significance.
• ClinVar API summary: classification 'Uncertain significance', review status 'criteria provided, multiple submitters, no conflicts' (2 stars) - an aggregate of laboratory submissions, not an expert-panel classification, and therefore incapable of supporting PP5.
• Generic ACMG/AMP 2015 PP5 definition (Richards et al. 2015, PMID 25741868) consulted as the governing framework text.

Sources

Reasoning

Under operating policy, BP6 is triggered only by an exact-variant ClinVar expert-panel classification of Benign or Likely benign. The ClinVar submission audit for NM_000077.5(CDKN2A):c.170C>G (p.Ala57Gly) (variation 187272) shows expert_panel_submissions = 0; the aggregate classification is Uncertain significance (7 clinical laboratories, 'criteria provided, multiple submitters, no conflicts', 2 stars). No expert-panel benign or likely-benign assertion exists for this variant, and no single-submitter laboratory asserts benign/likely benign (all 8 submissions are Uncertain significance). BP6 must never be triggered by ordinary laboratory submissions or aggregate ClinVar labels. BP6 is therefore not met.

Rule

BP6 (Richards et al. 2015), restricted by operating policy: reputable-source (ClinVar expert-panel) recent report of the exact variant as Benign/Likely benign, without independent laboratory evidence.

Evidence considered

• Operating policy for this pipeline: BP6 applies only for an exact-variant ClinVar expert-panel classification of Benign/Likely benign; ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions never trigger BP6.
• ClinVar submission audit (variation 187272): submissions_total=8, expert_panel_submissions=0, all submissions classified Uncertain significance by single-submitter clinical laboratories; no benign or likely-benign assertion from any source.
• ClinVar API summary: aggregate classification 'Uncertain significance' (2 stars, multiple submitters, no conflicts) - an aggregate laboratory label, which is explicitly excluded from triggering BP6.
• Generic ACMG/AMP 2015 BP6 definition (Richards et al. 2015, PMID 25741868) consulted as the governing framework text.

Sources

Reasoning

PS1 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

Generic ACMG/AMP 2015 (PMID 25741868) governs because no ClinGen CSPEC and no gene-specific VCEP materials exist for CDKN2A (vcep_materials.json: cspec found=false; framework_mode=generic_acmg_fallback). PS2 requires a de novo occurrence with confirmed maternity/paternity documented in the proband via trio testing. No proband-parent (trio) genotype data, parental-testing result, or de novo assertion for NM_000077.5:c.170C>G (p.Ala57Gly) appears anywhere in the case-evidence bundle: the ClinVar record 187272 submissions are all single-submitter Uncertain significance with no de novo claim captured (no expert panel), and none of the fetched full-text papers (paper_extracts.json) reports proband-parent genotyping or a de novo occurrence for this variant. The de novo evidence base is absent, so PS2 is not assessed rather than failed; no source in the registry supports it.

Rule

PS2: De novo occurrence with confirmed maternity and paternity (generic ACMG/AMP 2015, PMID 25741868).

Evidence considered

• Framework check: no ClinGen CSPEC or VCEP specification available for CDKN2A; generic ACMG/AMP 2015 standards (PMID 25741868) applied to PS2.
• PS2 rule consulted (PMID 25741868): de novo variant with confirmed maternity and paternity in the proband.
• Searched the consolidated bundle (ClinVar record 187272 and all 8 submissions, literature_pass.json criteria_index/criterion_screened, evidence.json, paper_extracts.json for all fetched papers) for any de novo, parental-testing, or proband-parent observation of c.170C>G; none present.
• ClinVar submissions for NM_000077.5(CDKN2A):c.170C>G (p.Ala57Gly) are all single-submitter Uncertain significance (7 clinical laboratories, no expert panel, 2 review stars); per policy, these final classifications were not used as evidence for or against PS2.

Still missing

• No proband-parent trio testing results or parental genotypes are reported in any collected source; de novo occurrence with confirmed parentage can neither be established nor excluded.
• ClinVar criterion-lead submissions (SCV002534312, SCV004212497, SCV000545525, SCV000217810, SCV005197432, SCV000903860, SCV001134101) were flagged as having PMID trails, but their full assertion text was not present in this bundle - a human reviewer with SCV access should confirm none of them asserted PS2.

Sources

Reasoning

No ClinGen CSPEC/VCEP or local custom gene framework exists for CDKN2A (vcep_materials.json: no cspec URL; final_classification_framework found=false), so generic ACMG/AMP 2015 (Richards et al., PMID:25741868) governs: PS3 requires well-established functional studies (validated in vitro/in vivo assays) demonstrating a deleterious effect of the variant. Every source with potential variant-level functional content was checked for the exact variant NM_000077.5:c.170C>G (p.Ala57Gly): (1) PMID:19260062, the dedicated CDKN2A germline functional study (in vitro CDK4 co-immunoprecipitation binding and primary HDF proliferation-arrest assays on 20 mutations), does not test A57G — full text searched, variant absent; its only codon-57 entry is the different substitution c.170C>T p.Ala57Val, which retained ~84% of WT CDK4 binding and was indistinguishable from WT in proliferation arrest. (2) PMID:21462282 lists p.A57G only as a clinico-epidemiologic VUS (Table 1C: 0 familial-melanoma families, 1 multiple-primary-melanoma carrier) with no U2OS cell-cycle arrest assay result reported for it. (3) OncoKB (A57G) reports no variant-specific reviewed functional evidence ('Unknown Oncogenic Effect', not_reviewed=true). (4) ClinVar submission SCV000903860 (Color) states 'to our knowledge, functional studies have not been reported for this variant', corroborating absence of published functional studies (used only as a factual statement, not as a criterion inference from another lab's classification). In-silico predictions (SpliceAI max delta 0.00; REVEL 0.205) are not functional assays and cannot satisfy PS3. No well-established functional study demonstrates a deleterious effect for this exact variant; evidence is insufficient, so PS3 is not assessed rather than met.

Rule

ACMG/AMP 2015 PS3 (PMID:25741868): well-established functional studies show a deleterious effect; generic ACMG fallback applies (no CDKN2A CSPEC/VCEP or local custom framework). Default strength when met: strong. Not assessed here because no variant-specific functional study exists.

Evidence considered

• Governing framework: generic ACMG/AMP 2015 fallback (Richards et al. PMID:25741868) because vcep_materials.json and final_classification_framework.json found no ClinGen CSPEC/VCEP or local custom CDKN2A specification (cspec found=false; framework_complete=false).
• PMID:19260062 full text searched for c.170C>G / p.(A57G) / A57G: the exact variant is not studied. Assays used were in vitro CDK4 coimmunoprecipitation binding (relative to WT p16INK4A=100%) and proliferation-arrest in primary human diploid fibroblasts (Hs68 HDFs). The only codon-57 variant tested was the different substitution Ala57Val (c.170C>T), which retained ~84% of WT CDK4 binding and was 'virtually indistinguishable from WT' in blocking HDF proliferation; the authors kept A57V as an unclassified variant due to discordant genetic data. No assay result of any kind exists for A57G in this paper.
• PMID:21462282 full text searched: p.A57G appears only in Table 1C ('CDKN2A Variants of Uncertain Significance', codon 57 row: 0 FM families, 1 MPM carrier, 1 total carrier). No functional (U2OS cell-cycle arrest), computational, or Bayesian result is reported for p.A57G; the paper's 'wild-type in vitro function' odds ratio (0.28) belongs to the same-codon p.A57V, and the paper explicitly cautions that different variants at the same codon can have different functional outcomes.
• OncoKB CDKN2A A57G page: classification 'Unknown Oncogenic Effect', no variant-specific reviewed functional evidence identified (not_reviewed=true); gene-level context only, so it provides no PS3-supporting functional data.
• ClinVar prefetch/audit (SCV000903860, Color Diagnostics) contains the statement 'to our knowledge, functional studies have not been reported for this variant' under a PS3-tagged functional_assay signal; treated strictly as corroboration that no functional studies exist, never as an independent criterion call.
• No other collected source (PMID:15146471, PMID:22447455, PMID:26467025, PMID:29316957, PMID:31672839, PMID:25741868, gnomAD, COSMIC, Cancer Hotspots, SpliceAI, REVEL) reports any functional assay result for c.170C>G p.(Ala57Gly).

Still missing

• No published well-established functional assay data (e.g., p16INK4A CDK4/CDK6 binding, kinase-inhibition, or cell-cycle arrest/senescence assay) exist for the exact variant c.170C>G p.(Ala57Gly) in the collected corpus; variant-specific functional testing would be required to apply PS3.
• CDKN2A exon 2 is translated in two alternative reading frames (p16INK4A and p14ARF); any future functional study should clarify which gene product/isoform is assayed and whether p14ARF function could be affected.
• Functional results reported for the same-codon substitution p.Ala57Val (PMID:19260062; PMID:21462282) cannot be extrapolated to p.Ala57Gly and do not substitute for PS3 evidence.

Sources

Reasoning

PM1 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

PM2 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

PM5 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

PM6 (assumed de novo without confirmation of parentage) requires at least an observed de novo event in the proband with parental testing absent or uninformative. No de novo occurrence of any kind is reported for c.170C>G in any collected source - no trio data, no parental testing results, and no submission or paper states the variant arose de novo. The single p.A57G carrier reported in the literature (PMID:21462282, one multiple-primary-melanoma carrier; PMID:22447455, one unaffected population control) has no parental genotype information attached. Evidence is insufficient; PM6 is not assessed.

Rule

PM6: Assumed de novo occurrence, without confirmation of parentage (generic ACMG/AMP 2015, PMID 25741868).

Evidence considered

• PM6 rule consulted (PMID 25741868): assumed de novo occurrence without confirmation of parentage.
• Paper extracts for every fetched full-text paper were checked for de novo or parental-testing content on c.170C>G/p.Ala57Gly; none reported.
• Literature and ClinVar sources that do mention p.Ala57Gly (PMID:21462282 Table 1C single MPM carrier; PMID:22447455 single control carrier) attach no parental genotype or de novo information.
• No VCEP/CSPEC modification of PM6 applies (generic ACMG/AMP fallback for CDKN2A).

Still missing

• No proband observation consistent with de novo origin exists in the collected evidence; PM6 cannot be applied in the absence of any parental testing information, informative or otherwise.
• If proband family history/testing data exist outside this bundle (e.g., in the originating laboratory requisition), they would be required to assess PM6.

Sources

Reasoning

PP1 requires co-segregation of the variant with disease in multiple affected family members (informative meioses). The only variant-level segregation-adjacent datum in the bundle is the CDKN2A germline-variant database listing in PMID:21462282 (Table 1C): p.A57G is tabulated with 0 familial-melanoma (FM) families, 1 multiple-primary-melanoma (MPM) carrier, 0 sporadic-melanoma carriers, 0 non-melanoma carriers (1 total carrier). This shows the variant was never observed in a familial-melanoma kindred in that database and supplies no informative meioses; it is an absence of positive segregation evidence, not a quantitative segregation result. With no affected relatives genotyped anywhere in the collected evidence, PP1 cannot be applied; the absence of demonstrated co-segregation must not be double-counted as BS4 (that criterion is assessed separately and likewise lacks data).

Rule

PP1: Co-segregation of the variant with disease in multiple affected family members (generic ACMG/AMP 2015, PMID 25741868).

Evidence considered

• PMID:21462282 Table 1C lists p.A57G among CDKN2A variants of uncertain significance: '57 p.A57G 0 1 0 0 1' = 0 familial-melanoma families, 1 MPM carrier, 0 sporadic-melanoma carriers, 0 non-melanoma carriers, 1 total carrier - no co-segregation with familial melanoma is reported for this variant.
• PP1 rule consulted (PMID 25741868): co-segregation of the variant with disease in affected family members; generic ACMG/AMP counts segregation data only when multiple affected relatives are genotyped.
• No other source in source_registry contains family or segregation data for c.170C>G; literature_pass.json lists PMID:22447455 under a 'segregation' theme label, but its full-text extraction shows it is a population case-control study of MC1R/ASIP/TYR/TYRP1 with no CDKN2A family segregation analysis.
• Database tabulation counts (PMID:21462282) are raw variant-level data, not another source's final classification, and were used only as direct evidence of absent segregation observations.

Still missing

• No affected relatives of any p.A57G carrier have been genotyped or reported; zero informative meioses available for segregation analysis.
• The CDKN2A database entry (PMID:21462282) reports no FM family carrying p.A57G, but this is registry-absence data without pedigree-level genotyping and cannot substitute for a segregation study.
• A human reviewer would need family-study data (e.g., melanoma-affected relatives genotyped for the variant) to assess PP1.

Sources

Reasoning

PP2 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

PP4 requires a phenotype or family history highly specific for a disease with a single genetic etiology. The case bundle contains no index-case phenotype or family-history information at all: case_summary.json has no proband phenotype section, and the only phenotypic strings available are ClinVar record-level condition labels (Familial melanoma, Hereditary cancer-predisposing syndrome, Melanoma and neural system tumor syndrome, not provided) attached to laboratory submissions, which are not a documented proband phenotype. In addition, melanoma/cancer predisposition is a common and genetically heterogeneous phenotype (multiple susceptibility genes), so phenotype specificity of PP4 grade could not be established even from gene-level context. With no phenotype data, PP4 cannot be adjudicated and is returned as not_assessed.

Rule

PP4 (Richards et al. 2015): phenotype or family history highly specific for a disease with a single genetic etiology.

Evidence considered

• Reviewed the consolidated case-evidence bundle (case_summary.json, evidence.json, clinvar submission audit): no proband phenotype or family-history description is recorded for this case.
• ClinVar variation 187272 carries condition labels (Familial melanoma; Hereditary cancer-predisposing syndrome; Melanoma and neural system tumor syndrome; not provided) from laboratory submissions; these are submitter-level assertions for the record, not a documented proband phenotype, and were not used to infer PP4.
• PP4 (Richards et al. 2015) requires a phenotype/family history highly specific for a disease with a single genetic etiology; cutaneous melanoma and cancer-predisposition phenotypes are common, phenocopy-prone, and genetically heterogeneous, which would in any event limit PP4 applicability without explicit single-etiology phenotype documentation.

Still missing

• No proband phenotype, personal cancer history, or family-history data is present in this case's evidence bundle; a documented phenotype highly specific for CDKN2A-related melanoma susceptibility would be required to evaluate PP4.

Reasoning

BA1 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

BS1 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

BS2 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

Generic ACMG/AMP 2015 (Richards et al., PMID:25741868) governs (no CDKN2A CSPEC/VCEP or local custom framework): BS3 requires well-established functional studies demonstrating no damaging effect on protein function or splicing for the variant in question. No such study exists for the exact variant c.170C>G (p.Ala57Gly). The nearest functional data concern a different substitution at the same codon: PMID:19260062 showed p.Ala57Val (c.170C>T) retains ~84% of WT CDK4 binding and behaves like WT in HDF proliferation arrest, and PMID:21462282 scored p.Ala57V as wild-type in its in vitro cell-cycle assay — but those results are for Ala57Val, not Ala57Gly, and PMID:21462282 explicitly cautions that different variants at the same codon can have different functional outcomes, so they cannot be extrapolated to establish BS3 for this variant. PMID:22447455 reports the exact variant p.Ala57Gly in 1 of 420 healthy blood-donor controls but performed no functional assay; a single healthy-control observation is population-based evidence (relevant to BS1/BA1-family criteria handled elsewhere), not a well-established functional study, and does not satisfy BS3. In-silico predictions (SpliceAI max delta 0.00; REVEL 0.205) are not functional assays and cannot satisfy BS3. Evidence is insufficient in the damaging-effect-negative direction, so BS3 is not assessed rather than met.

Rule

ACMG/AMP 2015 BS3 (PMID:25741868): well-established functional studies show no damaging effect on protein function or splicing; generic ACMG fallback applies (no CDKN2A CSPEC/VCEP or local custom framework). Default strength when met: strong. Not assessed here because no variant-specific functional study exists.

Evidence considered

• Governing framework: generic ACMG/AMP 2015 fallback (Richards et al. PMID:25741868) because no ClinGen CSPEC/VCEP or local custom CDKN2A specification is available (cspec found=false; vcep_materials.json empty; final_classification_framework found=false).
• PMID:19260062 (functional, structural, and genetic evaluation of 20 CDKN2A germline mutations) was reviewed in full text for a BS3-relevant result for A57G: none exists. Its near-WT assay results (84% CDK4 binding; HDF proliferation indistinguishable from WT) apply only to the same-codon, different-residue change p.Ala57Val (c.170C>T), which the authors still classified as an unclassified variant.
• PMID:21462282 was reviewed in full text: p.A57G has no functional laboratory result (Table 1C VUS listing only); the wild-type in vitro function score applies to p.A57V, and the paper cautions that different substitutions at the same codon can have different functional outcomes, precluding same-codon extrapolation for BS3.
• PMID:22447455 mentions the exact variant: 'One CDKN2A mutation (Ala57Gly) was detected in one subject of the control population' (1/420 blood donors). This is a carrier observation without any functional assay and without pathogenicity conclusion; it does not constitute a well-established functional study and therefore cannot support BS3.
• PMID:15146471 is a literature review of germline CDKN2A mutations; its codon-57 entry is c.170C>T p.A57V (1 French family), not c.170C>G, and it reports no functional data; reviewed and excluded as BS3 evidence.
• In-silico-only signals for this variant (SpliceAI max delta 0.00; REVEL 0.205) are predictor outputs, not well-established functional studies, and cannot satisfy BS3 under the generic ACMG/AMP framework.

Still missing

• No well-established functional study of the exact variant c.170C>G p.(Ala57Gly) demonstrates absence of a damaging effect; BS3 would require variant-specific assay data (e.g., CDK4/CDK6 binding or cell-cycle arrest comparable to WT controls).
• Same-codon p.Ala57Val near-WT functional results (PMID:19260062; PMID:21462282) are not transferable to p.Ala57Gly, and no validated assay has been reported for the Gly substitution.
• The observation of p.Ala57Gly in one healthy control (PMID:22447455) is not functional evidence and cannot be used for BS3; whether any population-frequency threshold applies is outside this group's criteria.

Sources

Reasoning

BS4 requires demonstrated lack of segregation in affected members of a family (e.g., an affected relative who does not carry the variant). No family studies, pedigree genotyping, or non-segregation observations for c.170C>G exist in the bundle. The closest observation, PMID:22447455, reports one Ala57Gly carrier among 420 healthy Norwegian blood-donor controls and does not list Ala57Gly among the CDKN2A mutations found in the 388 MPM patients - a single unaffected population-control carrier with no affected relatives tested. That is population-frequency/penetrance context (assessed under other groups' criteria such as BS2/BS1/BA1) and does not by itself establish lack of segregation within affected family members. Evidence is insufficient for BS4; the criterion is not assessed, and the control-carrier observation must not be inflated into BS4.

Rule

BS4: Lack of segregation in affected members of a family (generic ACMG/AMP 2015, PMID 25741868).

Evidence considered

• PMID:22447455 reports 'One CDKN2A mutation (Ala57Gly) was detected in one subject of the control population' (Materials and Methods, Recruitment of Patients and Controls) - i.e., 1/420 healthy blood-donor controls carries p.Ala57Gly; the paper draws no pathogenicity conclusion and provides no family data.
• BS4 rule consulted (PMID 25741868): lack of segregation in affected members of a family; a population-control carrier observation does not meet this definition because no affected relatives were tested.
• PMID:21462282 Table 1C shows no familial-melanoma family with p.A57G, but registry-level absence of affected-family observations was not treated as demonstrated non-segregation.
• No VCEP/CSPEC modification of BS4 applies (generic ACMG/AMP fallback for CDKN2A).

Still missing

• No affected family member of any p.A57G carrier has been genotyped as non-carrier; no pedigree segregation analysis exists in any collected source.
• A single unaffected population-control carrier (PMID:22447455, 1/420) is not non-segregation evidence and cannot support BS4 without affected-relative genotyping.
• Human review would be needed if family-testing records (affected relative non-carrier results) exist outside this bundle.

Sources

Reasoning

BP1 could not be assessed because its agent did not produce valid output.

Still missing

• attempt 1: Hermes timed out after 200s and left no usable output_file
• attempt 2: Hermes timed out after 200s and left no usable output_file
• attempt 3: Hermes timed out after 200s and left no usable output_file

Reasoning

Generic ACMG/AMP 2015 BP2 requires a documented observation of the variant either in trans with a pathogenic variant (for a fully penetrant dominant gene/disorder) or in cis with a pathogenic variant (any inheritance pattern). No such phase or co-occurrence observation exists anywhere in this case's evidence: all 8 ClinVar submissions (7 usable, all VUS, single-submitter, no expert panel) report the variant without any second CDKN2A variant or phase information; the fetched literature contains no familial genotyping of c.170C>G (PMID:21462282: 1 MPM carrier, 0 FM families, no segregation; PMID:22447455: 1 carrier among 420 control blood donors). Evidence is therefore insufficient to assess BP2 rather than contradictory. Additionally, CDKN2A melanoma predisposition is dominant with incomplete penetrance, which limits (but the cis clause leaves technically applicable) the trans arm of BP2; no cis observation was reported either. Phase testing of the proband/family or a documented co-occurrence with a pathogenic CDKN2A variant would be required before BP2 could be evaluated.

Rule

Generic ACMG/AMP 2015 (Richards et al., PMID 25741868): BP2 - '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' (supporting benign). Not assessable: no phase/co-occurrence observation available.

Evidence considered

• Governing framework: no ClinGen CSPEC/VCEP specification exists for CDKN2A, so the generic ACMG/AMP 2015 BP2 definition (PMID 25741868) applies: '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' (supporting benign).
• ClinVar variation 187272 (NM_000077.5:c.170C>G, p.Ala57Gly): 8 submissions, 7 usable for classification, all 'Uncertain significance' with criteria provided by a single submitter and no expert-panel input; the submission audit contains no co-occurrence, phase, or second-variant statements for any submission.
• PMID:21462282 (Table 1C) is the only affected-proband data source for p.A57G: 0 familial-melanoma families, 1 multiple-primary-melanoma carrier, 1 total carrier; no segregation with familial melanoma and no phase/cis-trans information is reported for this variant.
• PMID:22447455 reports Ala57Gly in 1 of 420 anonymous Norwegian blood-donor controls (population-based series of multiple primary melanomas); the carrier is heterozygous with no second CDKN2A variant, phenotype, or phase data reported.
• Disease-context caveat: CDKN2A melanoma susceptibility is dominant with incomplete penetrance, so the BP2 trans arm (written for fully penetrant dominant disorders) would be of limited weight even if a trans observation existed; the cis arm is applicable in principle but no cis observation was found.

Still missing

• No cis/trans phase observation of c.170C>G with any pathogenic CDKN2A variant is documented in ClinVar submissions or in the fetched full-text literature.
• No familial genotyping, segregation, or co-occurrence data for the proband/case was available in the evidence bundle (no second CDKN2A variant reported in any carrier).

Sources

Reasoning

BP5 requires the variant to be found in a case with an alternate molecular basis for disease (e.g., a pathogenic variant in a different gene that explains the phenotype). The case bundle contains no proband-level testing context whatsoever: there is no documented index case, no alternate pathogenic finding, and no dual-diagnosis or co-occurrence information that would indicate the phenotype is explained by another molecular cause. With no case-level data, BP5 cannot be evaluated and is returned as not_assessed. (Observations of this variant in unaffected population controls - 1/420 blood donors in PMID:22447455 and low-frequency gnomAD presence - relate to benign population-frequency criteria such as BS1/BS2, not to BP5, and were not used here.)

Rule

BP5 (Richards et al. 2015): variant found in a case with an alternate molecular basis for disease.

Evidence considered

• Reviewed the consolidated case-evidence bundle for any case-level alternate-molecular-basis evidence (co-occurrence, dual diagnosis, alternative genetic explanation): none is present; the bundle contains no proband phenotype or testing context.
• BP5 (Richards et al. 2015) definition consulted: 'Variant found in a case with an alternate molecular basis for disease'; without a documented case with an alternate cause, the criterion cannot be assessed.
• Noted for disambiguation: population/control observations of the variant (PMID:22447455 control carrier; gnomAD v2.1 9/251,530 and v4.1 22/1,597,598 alleles) are population-frequency evidence evaluated under BS1/BS2/BA1/PM2 by their own group, not BP5 evidence.

Still missing

• No proband-level data describing an alternate molecular basis for the phenotype (e.g., co-occurrence with a pathogenic variant in another gene that fully explains the clinical presentation) is available in the evidence bundle; BP5 requires such a documented case context.