Neural networks for analysis of single cell RNAseq and amino acid sequences have been developed

Novel Software Systems LLC participated in Moscow Conference on Computational Molecular Biology (MCCMB) at the Skolkovo Institute of Science and Technology, from July, 30 to August, 2, 2021.

At the conference our company was represented by:

  • Denis Antonec, PhD in Biology who delivered a presentation “Style transfer with variational autoencoders is a promising approach to RNA-Seq data harmonization and analysis”. It was devoted to a method based on in-depth training, which allows to simulate a change in gene expression profile in response to drug therapy.
  • Nikolay Russkikh, Head of Machine Learning department, who gave a presentation entitled “Functional domain annotation of protein sequences with deep metric learning”. His presentation focused on the method of predicting functional domains in amino acid sequences by analyzing small-dimensional representations obtained through deep learning methods. This method addresses the so-called long-range homology problem and allows the analysis of a wide variety of protein sequences, including viruses without homologues with a known function.

The technologies presented are used to develop new diagnostic and therapeutic techniques as well as to design substances with determined properties.