Deciphering exome sequencing data: Bringing mitochondrial DNA variants to light

The expanding use of exome sequencing (ES) in diagnosis generates a huge amount of data, including untargeted mitochondrial DNA (mtDNA) sequences. We developed a strategy to deeply study ES data, focusing on the mtDNA genome on a large unspecific cohort to increase diagnostic yield. A targeted bioinformatics pipeline assembled mitochondrial genome from ES data to detect pathogenic mtDNA variants in parallel with the "in-house" nuclear exome pipeline. mtDNA data coming from off-target sequences (indirect sequencing) were extracted from the BAM files in 928 individuals with developmental and/or neurological anomalies. The mtDNA variants were filtered out based on database information, cohort frequencies, haplogroups and protein consequences. Two homoplasmic pathogenic variants (m.9035T>C and m.11778G>A) were identified in 2 out of 928 unrelated individuals (0.2%): the m.9035T>C (MT-ATP6) variant in a female with ataxia and the m.11778G>A (MT-ND4) variant in a male with a complex mosaic disorder and a severe ophthalmological phenotype, uncovering undiagnosed Leber\u27s hereditary optic neuropathy (LHON). Seven secondary findings were also found, predisposing to deafness or LHON, in 7 out of 928 individuals (0.75%). This study demonstrates the usefulness of including a targeted strategy in ES pipeline to detect mtDNA variants, improving results in diagnosis and research, without resampling patients and performing targeted mtDNA strategies