MAGE enables population level RNAseq driven genotyping and (differential) allelic divergence detection in healthy kidney and carcinoma
Stroobandt, C.; Goovaerts, T.; Coussement, L.; De Graeve, F.; Voorthuijzen, F.; Van Steenbergen, L.; Galle, J.; Van Criekinge, W.; De Meyer, T.
Show abstract
Decreasing sequencing costs have instigated large-scale RNAseq experiments, yet genetic polymorphisms in such data remain poorly exploited. Currently, allele-specific expression (ASE) studies focus almost exclusively on genetic variants explaining expression differences (cis-eQTLs), largely ignoring other ASE effects. The latter are typically associated with higher variance in expression of both copies of a gene, here called Allelic Divergence (AD). We therefore developed an RNAseq-driven population-level beta-binomial mixture model for (differential) AD detection. The model simultaneously enables RNAseq-driven genotyping, which outperforms alternative RNA genotyping methods when applied on healthy kidney data from The Cancer Genome Atlas. Moreover, we identify well-known non-cis-eQTL ASE, e.g. random monoallelic expression of HLA and immunoglobulin genes in healthy kidney, as well as allele-specific aberrations in clear cell kidney carcinoma, including long-range epigenetic silencing of protocadherins, copy-number alterations, and loss of imprinting. These methods are available as the Modeller of Allelic Gene Expression (MAGE) tool suite: https://biobix.github.io/MAGE/.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Predicting gene expression from histone marks using chromatin deep learning models depends on histone mark function, regulatory distance and cellular states 95%
- NetActivity enhances transcriptional signals by combining gene expression into robust gene set activity scores through interpretable autoencoders 95%
- Deep integrative models for large-scale human genomics 94%
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.