Chatomics Field GuideWhat They Don't Teach You

Assay

Variant Calling: sanity checks and pitfalls

Reads are aligned to a reference build, duplicates are marked, base quality scores may be recalibrated, and variants are called per sample or jointly across a cohort (GATK HaplotypeCaller, DeepVariant, bcftools). Then filtering, annotation and interpretation, where most of the judgment lives. The classic mistakes are mixing hg19 and hg38 coordinates, using a reference whose contig names do not match the annotation, trusting variants in low-mappability regions, and never running a sex or relatedness check that would have caught a sample swap.

Who this is for: Clinical and translational genomics analysts producing germline or somatic VCFs from exome or genome data, often under a pipeline they inherited. They need to know which variants are real, which are artifacts of alignment or mapping, and whether the sample they sequenced is the sample on the label.