Figure 2

Variant calling scheme for classical SNP and InDel caller and RAREVATOR. Panels a and b report the calling scheme of classical germline and somatic variant callers. Panels c and d report the calling scheme of RAREVATOR for germline and somatic variants. Classical germline callers (panel a) are devised to detect variants from the comparison to the reference genome and, clearly, they are not able to detect variants in RRA loci when the reference allele is in homozygous state. Also heterozygous calls are missed by classical resequencing analysis pipelines, since variant calling is usually followed by a filtering step in which known variants with allele frequency larger than a predefined threshold (usually 0.01) are discarded. Somatic callers (panel b) are devised to detect every kind of differences between normal and tumor samples. However, the probabilistic calling step is followed by a filtering step in which germline variants found in normal sample are used to discard somatic calls. RAREVATOR exploits the GATK Unified Genotyper for interrogating all the RRA loci and detecting germline and somatic variants that contain the reference allele (panels c and d).