AIRI Institute of Artificial Intelligence experts, together with colleagues from the Moscow Institute of Physics and Technology, developed the RMT architecture – Repetitive Memory Transformer – for the GENA neural network, previously developed in Russia. Olga Kardymon, a bioinformatician, researcher and head of the AIRI Bioinformatics group, told socialbites.ca.
The GENA neural network can analyze DNA sequences and look for patterns in them. It can be used for determining the effect of mutations on the operation of genes, searching different parts of the genome, classifying living organisms based on sequencing data, and other tasks.
“The first version of GENA runs about 3,000 nucleotides long, while the second architecture of the model allows 24,000 nucleotides to be fed into the input. Both models are already public and available to the world community. We have our own new proprietary architecture called RMT – Repetitive Memory Transformer, which was developed at AIRI together with our colleagues from the Moscow Institute of Physics and Technology. Kardymon can work with potentially unlimited lengths of text strings and a range of tasks where it successfully copes with 1-2 million letter lengths.
According to the expert, increasing the length of the DNA sequence analyzed by the neural network is an extremely important task for scientists.
“Because you need to understand the context where the mutations happen. The longer we give the sequence, the more the neural network can see and learn some patterns. Our task is to increase the array that the neural network can analyze,” the bioinformatician added.
To learn more about how neural networks look for mutations in the human genome, create proteins that did not exist in nature before, and predict the effectiveness of vaccines and drugs, see material “socialbites.ca”.