The robust bioinformatic analysis of the protein sequences with phase behavior

Liquid-liquid phase separation (LLPS) of many proteins is critical in the biological function of membrane-less organelles. Reversible nature of biomolecular PS in cells suggests that phases of proteins and nucleic acids are like to be tightly regulated. Post-translational modifications and single-residue mutations have been shown to lead to the dissolution of biomolecular droplets or phase transition into aggregated forms. Based on the wealth of experimental data, it reasonable to expect that the PS of proteins is highly sequence-specific.
To reveal common motifs of protein sequences that promote PS, we have carried out large-scale statistical analysis of phase separating protein deposited in the LLPSDB database. For each motif, the most significant interactions have been identified (e.g. charge, hydropathy, polarity, pi-stacking). The bioinformatic analysis provides us with crucial knowledge about sequence conservation, secondary structure preferences, aggregation tendency, post-translational modification prediction, and binding affinity.
The long-term goal of our research is the creation of a publically accessible database that would be used by both experimentalists for designing controlled mutations and for computationally-oriented scientists for developing new modeling tools.