Predicting heteropolymer phase separation using two-chain contact maps.

Area of Science:

• Biophysics
• Polymer Science
• Computational Biology

Background:

• Phase separation in polymer solutions is typically linked to single-chain (radius of gyration, Rg) and two-chain (second virial coefficient, B22) properties.
• However, Rg and B22 can be insufficient for distinguishing phase-separating from non-phase-separating heteropolymers, such as intrinsically disordered proteins (IDPs).

Purpose of the Study:

• To develop a predictive approach for heteropolymer phase separation using two-chain simulation data.
• To compare the efficacy of contact map analysis against traditional metrics (Rg, B22) for predicting phase separation.

Main Methods:

• Utilized two-chain simulations to generate contact maps, detailing monomer proximity between polymer chains.
• Trained phase-separation classifiers using statistical properties derived from contact maps for both minimal heteropolymer and residue-level IDP models.
• Compared the predictive accuracy of contact map-based classifiers with those based solely on Rg and B22.

Main Results:

• Simple statistical features of two-chain contact maps demonstrated high accuracy in predicting phase separation.
• Contact map analysis significantly outperformed classifiers relying on Rg and B22 alone.
• The developed method proved transferable across different heteropolymer models, including IDPs.

Conclusions:

• Two-chain contact maps provide a more accurate and spatially resolved understanding of heteropolymer phase separation than traditional metrics.
• This approach offers a computationally efficient and transferable method for identifying the driving forces behind IDP phase behavior in dilute solutions.