Keywords: graphical models, protein design, constraint satisfaction, computational protein design, AND/OR search, bucket elimination, mini-bucket elimination, weighted mini-bucket elimination, heuristic search, marginal MAP, branch-and-bound, cpd, computational biology, dynamic heuristics, cost shifting, ufo
TL;DR: This work presents several improvements upon existing computational protein re-design algorithm, AOBB-K*, noticeably improving scalability and leading to schemes that can be generalized to other well known tasks over graphical models.
Abstract: Scientific computing has experienced a surge empowered by advancements in technologies such as neural networks. However, certain important tasks are less amenable to these technologies, benefiting from innovations to traditional inference schemes. One such task is protein re-design. Recently a new re-design algorithm, AOBB-K*, was introduced and was competitive with state-of-the-art BBK* on small protein re-design problems. However, AOBB-K* did not scale well. In this work, we focus on scaling up AOBB-K* and introduce three new versions: AOBB-K*-b (boosted), AOBB-K*-DH (with dynamic heuristics), and AOBB-K*-UFO (with underflow optimization) that significantly enhance scalability.
Other Supplementary Material: zip