ABMGG Molecular Genetic Pathology common mistakes
ABMGG Molecular Genetic Pathology · Understand the traps and how to avoid them.
When interpreting a germline variant of uncertain significance (VUS), the candidate classifies it as pathogenic based on a single in silico prediction (e.g., SIFT or PolyPhen) without integrating population frequency, segregation, functional data, or ACMG/AMP criteria.
Candidates often over-rely on computational tools because they are easy to run and produce a numeric score, forgetting that ACMG/AMP guidelines require a weight-of-evidence approach and that in silico predictions alone are supporting evidence (PP3) and insufficient for a pathogenic classification.
What to do instead
Practice applying the ACMG/AMP 2015 framework (and ClinGen updates) to variants: explicitly assign evidence codes (e.g., PVS1, PS1, PM2, PP3) and require a combination that meets pathogenic/likely pathogenic rules. Review cases where in silico predictions conflict with functional or segregation data.
In somatic variant interpretation, the candidate reports a variant at a low variant allele fraction (VAF) as a true somatic mutation without considering tumor purity, stromal contamination, sequencing artifacts, or clonal hematopoiesis.
Candidates often focus on the variant call itself and forget that VAF is a composite of tumor cellularity, copy number, and technical noise; they may also overlook that low-VAF variants in blood or normal tissue can represent clonal hematopoiesis of indeterminate potential (CHIP).
What to do instead
Always estimate tumor purity from histology or orthogonal data, use paired normal controls when available, and apply domain-specific VAF thresholds. For hematologic or liquid biopsy assays, review CHIP-associated genes (DNMT3A, TET2, ASXL1) and consider orthogonal validation for low-VAF calls.
When designing a PCR-based assay, the candidate ignores primer-dimer potential, GC content, melting temperature mismatch, or the presence of pseudogenes, leading to nonspecific amplification or allele dropout.
Candidates memorize the basic steps of PCR but fail to apply thermodynamic principles and specificity checks (e.g., BLAST for pseudogenes, in silico PCR) to real primer sequences, especially for GC-rich or repetitive regions.
What to do instead
Practice using primer design software (Primer3, Primer-BLAST) and manually check: Tm difference ≤5°C, GC content 40–60%, avoidance of 3′ complementarity, and specificity against the reference genome including pseudogenes. Include a no-template control and a positive control with known heterozygous variant to detect allele dropout.
In quality assurance and validation, the candidate accepts a limit of detection (LOD) stated by the manufacturer without verifying it in-house using appropriate reference materials and replicates.
Candidates confuse FDA clearance or manufacturer claims with laboratory-specific validation requirements under CLIA and CAP, and they underestimate the need to establish LOD, precision, and accuracy for the laboratory’s own workflow, including extraction and instrument differences.
What to do instead
Review CLIA/CAP checklist requirements for validation of molecular assays: establish LOD with ≥20 replicates at the claimed LOD and ≥20 at a level near the LOD, test accuracy with ≥20 positive and ≥20 negative samples (or appropriate reference materials), and document precision. Use cell line dilutions or synthetic controls to verify LOD independently.
When interpreting a chromosomal microarray (CMA) result, the candidate reports a copy number variant (CNV) as pathogenic solely because it is large or contains genes, without considering population frequency, inheritance, and known benign CNV databases.
Candidates often equate size with pathogenicity and neglect to check the CNV against databases such as DGV, ClinGen, and DECIPHER, or to consider that many large CNVs are benign polymorphisms.
What to do instead
For every CNV, document: size, genomic coordinates (GRCh37/38), gene content, inheritance (if parental samples available), and frequency in population databases. Apply ACMG/ClinGen CNV interpretation guidelines (e.g., scoring for haploinsufficiency, gene count). Classify as VUS when evidence is insufficient.
In pharmacogenomics, the candidate interprets a genotype without translating it to a phenotype (e.g., poor metabolizer, intermediate metabolizer) and without considering the specific drug’s FDA/CPIC guidelines.
Candidates may focus on the genotype call (e.g., CYP2D6 *4/*4) but forget that clinical actionability depends on the drug, the phenotype, and the guideline (CPIC, DPWG, FDA). They may also overlook that star alleles require phased interpretation and that some drugs have multiple gene interactions.
What to do instead
Practice using CPIC guidelines: for each gene-drug pair, map genotype to phenotype using the CPIC term (e.g., CYP2D6 poor metabolizer) and then to the recommended action. Review common genes (CYP2D6, CYP2C19, CYP2C9, VKORC1, TPMT, DPYD, HLA-B*57:01) and their star allele definitions.
In hematologic malignancies, the candidate fails to integrate cytogenetic and molecular findings (e.g., BCR-ABL1, PML-RARA, NPM1, FLT3-ITD) with the WHO classification and risk stratification, instead reporting each abnormality in isolation.
Candidates often memorize individual mutations but do not practice synthesizing them into a diagnostic entity (e.g., AML with mutated NPM1) and a risk category (e.g., ELN risk), which is essential for clinical management.
What to do instead
Study the WHO 5th edition and ICC classifications and the ELN risk stratification for AML, ALL, and MDS. For each case, list all molecular and cytogenetic findings, then determine the integrated diagnosis and risk group. Use case-based practice with real reports.
In infectious disease molecular diagnostics, the candidate interprets a positive PCR result as proof of active infection without considering colonization, contamination, or the limits of detection, and without correlating with clinical context.
Candidates may treat molecular tests as definitive and forget that PCR detects nucleic acid, not viability, and that contamination (amplicon, reagent) or colonization can yield false positives. They may also ignore the need for confirmatory testing or quantitative results.
What to do instead
Always interpret molecular infectious disease results in the context of pre-analytic factors, assay limitations, and clinical presentation. Review the difference between qualitative and quantitative assays, the role of confirmatory testing (e.g., sequencing for resistance), and the use of appropriate controls (negative, positive, internal amplification control).
In bioinformatics and genomic medicine, the candidate ignores the reference genome build (GRCh37 vs GRCh38) and transcript nomenclature (e.g., NM_ accession, version) when reporting variants, leading to incorrect HGVS nomenclature and clinical interpretation.
Candidates often focus on the variant call and forget that genomic coordinates and transcript references are essential for unambiguous reporting; using the wrong build or an outdated transcript can shift a variant to a different codon or exon.
What to do instead
Always record the reference build and transcript accession with version number. Practice converting variants between builds using tools like UCSC LiftOver and verifying HGVS nomenclature with Mutalyzer or the HGVS website. For clinical reporting, use the MANE Select transcript when available.
In identity testing, the candidate interprets a mismatch at a single STR locus as proof of non-identity or non-relationship without considering mutation rates, allelic dropout, or the statistical weight of the full profile.
Candidates may over-interpret single-locus discrepancies and forget that STR mutations occur (especially in parentage testing) and that a single mismatch can arise from technical artifacts; they may also neglect to calculate a likelihood ratio or random match probability.
What to do instead
For identity testing, always evaluate the full profile, use validated STR kits, and apply statistical interpretation (e.g., likelihood ratio, combined probability of inclusion). For parentage, consider mutation models and test additional loci when a single mismatch is observed. Review CAP/CLIA requirements for identity testing.