Library preparation and sequencing
Genomic libraries of whole-exome were constructed using 1 ug of genomic DNA and the Sureselect QXT V6 (Agilent Technologies - Patients 1 and 2, and Patient 1’s mother), OneSeq Constitutional Research Panel (Agilent Technologies - Patient 2’s father), and xGen Exome Research Panel v1.0 (IDT - Integrated DNA Technologies - FFPE tumor sample) kits. Enriched libraries were sequenced on the Illumina HiSeq 2500 platform in paired end reads. The sequences were aligned to the GRCh37/hg19 human genome reference with the BWA_MEM algorithm 30. Picard tools (v.1.8, http://broadinstitute.github.io/picard/) were used to convert the SAM file into BAM and to mark PCR duplicates. The Genome Analysis Toolkit (GATK 3.7) 31 were used to realign indels, recalibrate the bases, and to call (Unified Genotyper) and recalibrate variants (VQSR). Finally, multiallelic variants were split into different lines using the script split_multiallelic_rows.rb from Atlas2 32 to obtain the VCF files used for analysis.