Peng, G. et al. Proc. Natl. Acad. Sci. USA 110, 3985–3990 (2013).

Identifying which variants are real among next-generation sequencing data is a challenge. Filtering strategies can remove a large fraction of false variant calls that are due to sequencing and alignment errors, but they also tend to remove true rare variants that are important for genetic studies. Peng et al. use sequencing data from families to increase the sensitivity for detecting rare variants. Their family-based sequencing program, or FamSeq, uses raw sequence data from all family members jointly to determine the level of support for each variant. This Bayesian network approach uses Mendelian transmission information and reduces false negatives by up to 33% in extended HapMap pedigrees.