ABPath Molecular Genetic Pathology common mistakes
ABPath Molecular Genetic Pathology · Understand the traps and how to avoid them.
Reading a variant description such as c.76A>C or c.88+1G>T and interpreting the position as an amino acid number or exon number rather than a coding-DNA or intronic coordinate.
Candidates memorize the general HGVS layout but never drill the distinction between c. (coding DNA, +1 = first intronic base) and p. (protein) coordinates, so the numbering system is recognized but not decoded.
What to do instead
Practice translating 20 variants back and forth between c. and p. nomenclatures, explicitly stating for each what the number refers to and where the intron boundary lies.
Classifying a variant as pathogenic because it is rare or absent from population databases, without applying ACMG/AMP evidence codes.
Rarity feels intuitively decisive, and candidates conflate 'absent from gnomAD' (PM2, a supporting-level criterion) with proof of pathogenicity.
What to do instead
For every classification exercise, write out the specific codes invoked (PVS1, PS1, PM2, PP3, BS1, etc.) and the strength assigned, and confirm a combination that actually reaches Likely Pathogenic.
Choosing an assay platform (Sanger, allele-specific PCR, MLPA, Southern blot, repeat-primed PCR) before determining what the variant class requires.
Platforms are studied as a list of features rather than as answers to specific analytical problems, so the reasoning step from variant type to method is skipped.
What to do instead
Build a two-column table mapping variant classes (SNV, small indel, large deletion/duplication, trinucleotide expansion, methylation defect, fusion) to the method that detects each and the method's blind spots.
Computing recurrence risk for an X-linked or mitochondrial pedigree using autosomal recessive logic, or ignoring reduced penetrance and gonadal mosaicism when a pedigree appears 'sporadic'.
Bayesian risk problems are usually practiced on classic AR/AD pedigrees, and atypical inheritance patterns are under-rehearsed.
What to do instead
Work through pedigrees in a fixed sequence: confirm inheritance pattern, then apply carrier probability, penetrance, and de novo/gonadal mosaicism as separate multipliers before calculating.
Missing the need for a separate validation or a documented limitation when a test is offered for a new specimen type, a new variant class, or an off-label purpose.
Method validation is studied as a one-time event at assay launch rather than as a scope that must match intended use.
What to do instead
For each scenario, state the intended use in one sentence and then ask whether the existing validation data cover that specimen type, variant class, and clinical claim.
Treating a single negative or low-level result in a neoplastic specimen as excluding a mutation, without accounting for tumor cellularity, limit of detection, or clonal heterogeneity.
Sensitivity is remembered as an abstract assay statistic rather than as a specimen-specific variable.
What to do instead
Before interpreting any tumor result, estimate neoplastic fraction and compare it to the assay's validated limit of detection; state explicitly when a negative is uninformative.
Reporting a pharmacogenomic genotype as a diplotype or phenotype without applying the current star-allele definition and activity score, or without noting the affected drug and guideline source.
Candidates learn a few high-profile drug-gene pairs but not the star-allele nomenclature system that underlies genotype-to-phenotype translation.
What to do instead
Practice converting raw variants (e.g., a defined SNV plus a copy-number change) into a star allele, then into an activity score and phenotype category, citing the consensus guideline.
In identity, histocompatibility, or forensic questions, reporting a match or exclusion without stating the statistical framework (random match probability, likelihood ratio, or HLA mismatch level) and the population data behind it.
The laboratory work is familiar, but the interpretive statistics are treated as an afterthought rather than as the reportable conclusion.
What to do instead
For every identity or HLA scenario, write the conclusion as a probability statement with the reference population and the assumptions (relatedness, population substructure) made explicit.
Confusing analytical sensitivity and specificity with clinical sensitivity and specificity when evaluating a screening or diagnostic assay, especially for low-prevalence conditions.
The four terms are learned together and the predictive-value consequences of prevalence are not practiced numerically.
What to do instead
Recompute a 2x2 table at two different prevalences for the same assay and note how positive predictive value changes while sensitivity stays fixed.
Overlooking a required regulatory or ethical step, such as documented informed consent for a secondary finding, a required disclaimer for a research-use or laboratory-developed test, or a data-privacy obligation for genomic data sharing.
Regulatory content is memorized as statute names rather than as decisions embedded in the testing workflow.
What to do instead
For each test category, list the consent, reporting, retention, and data-sharing obligations as a checklist and apply it to a short scenario rather than recalling the regulation in isolation.