Politecnico di Torino - Corso Duca degli Abruzzi, 24 - 10129 Torino, ITALY

+39 011 090 6100 info@tech-share.it

Computational method for biological sequences optimization

Antibodies affinity maturationArtificial enzymescovid-19Drug DiscoveryProtein designProtein optimization

Introduction

The design of new proteins, such as antibodies and enzymes, is subject of great interest in chemical and pharmaceutical research, with an expanding market of products and services. This patent concerns an innovative machine learning method  for the generation of protein sequences optimized for a specific biochemical function.

Technical features

The prediction of protein sequences with a specific functionality is a problem that has enormous repercussions on the pharmaceutical, biomedical and industrial fields.One of the most used approaches is the Directed Evolution, which makes it possible to characterize the functionality of combinatorial libraries of mutants with the aim of selecting the variants with an increased functionality of interest. By combining these experiments with recent DNA sequencing techniques, it is possible to obtain a detailed mapping of the genotype-phenotype association of the combinatorial library.The present invention uses an innovative machine-learning method which, starting from library sequencing, produces an accurate statistical modeling of this association, allowing the construction of models for the generation of new amino acid sequences with biochemical functions of interest.

Possible Applications

  • Drug discovery;
  • Antibody affinity maturation;
  • Optimization Industrial Enzymes (sector: textiles, foods, detergents);
  • Optimization of affinity, stability, specific activity, solubility and others.

Advantages

  • In silico generation of new sequences;
  • Excellent precision in scoring the targeted biological function;
  • Applicable to different experimental schemes;
  • Can be integrated with other approaches;
  • Machine Learning assisted direct evolution;
  • Incoming data obtained with low-cost experiments.