How would you distinguish biological variants from sequencing or mapping artifacts?

Why

A good review separates true biological variation from technological explanations. First examine the site and neighboring region with the appropriate reference assembly. Mapping near paralogs, short repeats and indels can create apparently convincing artifacts.

Evaluate the read-level base and mapping qualities, allele distribution across strands and read positions, whether multiple reads came from one PCR molecule and whether the allele fraction is plausible for the assay and specimen. Compare known run artifacts, batch patterns and sample identity or mixture estimates. Apply the validated variant caller and filtering criteria, while recognizing that callers differ for germline, somatic and structural variants.

For example, a low-frequency allele seen almost exclusively at the ends of duplicates on one strand is less convincing than support from many independent molecules. Orthogonal confirmation may be appropriate when decision stakes and assay validation require it. Explicitly report poorly covered regions and uncertain findings rather than forcing every site into present or absent.