All research

Computational genomics & reproducible analysis · 2023

Variational autoencoders for RNA-seq outlier detection in Mendelian disease

Using RNA sequencing and variational autoencoders to find genes with unusual expression that may help explain rare Mendelian disorders.

Qatar Computing Research Institute, Hamad Bin Khalifa University · Research internship

Approach

Can unusual gene-expression patterns help prioritise candidate genes in Mendelian disease?

DNA sequencing does not explain every suspected Mendelian disorder. RNA sequencing adds functional evidence by showing when a gene is expressed far from its expected level. This internship project prepared RNA-seq data, used variational autoencoders to learn expected expression patterns, and examined outlier genes as candidates for follow-up.

  1. Prepare RNA-seq data
  2. Model expected expression
  3. Prioritise outlier genes

My contribution

I worked on RNA-seq preprocessing, variational autoencoder evaluation, anomaly analysis, gene prioritisation, interpretation of candidate genes and reproducible documentation.

Study scope

This was exploratory internship research and has not been published. Candidate genes would need independent biological and clinical validation before supporting a diagnosis.