Using Structural Protein Features for Long- and Medium-Range Contact Prediction
Mar 18, 2019
·
1 min read
Based on the idea that contact-maps of different proteins may in part be similar, we are evaluating the MaDCaT library as a search tool for improving predicted contact-maps. Furthermore, we evaluate the robustness of the method regarding noisy native contact-maps.
Authors
Heye Vöcking
Senior Data Engineer
Data & Knowledge Engineer with 10+ years of professional experience transforming petabyte-scale data into knowledge. Currently stress-testing large-language-model alignment, developing jailbreaks, and building real-time knowledge-graph systems. Interests include ML security, physics, Austrian economics, and Bitcoin.