Interpreting variants from whole-exome sequencing remains a major challenge, particularly for heterogeneous disorders such as mitochondrial diseases (MDs). To address this, we have developed Variant prIoritizatiOn using Latent spAce (VIOLA), a pipeline designed to help find a diagnosis for complex cases. VIOLA uses a variational autoencoder to embed functional annotations into a low-dimensional space, followed by DBSCAN-based outlier detection to identify potential pathogenic variants. Filtering steps and phenotype integration via HPO terms are then applied. The VIOLA score (Vscore) combines variant outlierness, transcriptomic co-expression data, and MD-specific annotations. Two rankings are derived: the VIOLA rank (all variants) and the ARrank (variants compatible with autosomal recessive inheritance). The VIOLA Aggregated score (VAscore) merges Vscore with Exomiser's pathogenicity score. Applied to 20 patients (four diagnosed), VIOLA reduced the variant list by >99% and ranked causal variants within the top 5 using ARrank, outperforming existing methods. Overall, VIOLA is a patient-specific strategy for variant prioritization, helping to resolve challenging MD cases and uncover novel disease mechanisms.

MOTIVATION: Current advances in omics technologies are paving the diagnosis of rare diseases proposing a complementary assay to identify the responsible gene. The use of transcriptomic data to identify aberrant gene expression (AGE) has demonstrated to yield potential pathogenic events. However, popular approaches for AGE identification are limited by the use of statistical tests that imply the choice of arbitrary cut-off for significance assessment and the availability of several replicates not always possible in clinical contexts. RESULTS: Hence, we developed ABerrant Expression Identification empLoying machine LEarning from sequencing data (ABEILLE) a variational autoencoder (VAE)-based method for the identification of AGEs from the analysis of RNA-seq data without the need for replicates or a control group. ABEILLE combines the use of a VAE, able to model any data without specific assumptions on their distribution, and a decision tree to classify genes as AGE or non-AGE. An anomaly score is associated with each gene in order to stratify AGE by the severity of aberration. We tested ABEILLE on a semi-synthetic and an experimental dataset demonstrating the importance of the flexibility of the VAE configuration to identify potential pathogenic candidates. AVAILABILITY AND IMPLEMENTATION: ABEILLE source code is freely available at: https://github.com/UCA-MSI/ABEILLE. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.