IgCraft: A versatile sequence generation framework for antibody discovery and engineering

Designing antibody sequences to better resemble those observed in natural human repertoires is a key challenge in biologics development. We introduce IgCraft: a multi-purpose model for paired human antibody sequence generation, built on Bayesian Flow Networks. IgCraft presents one of the first unified generative modeling frameworks capable of addressing multiple antibody sequence design tasks with a single model, including unconditional sampling, sequence inpainting, inverse folding, and CDR motif scaffolding. Our approach achieves competitive results across the full spectrum of these tasks while constraining generation to the space of human antibody sequences, exhibiting particular strengths in CDR motif scaffolding (grafting) where we achieve state-of-the-art performance in terms of humanness and preservation of structural properties. By integrating previously separate tasks into a single scalable generative model, IgCraft provides a versatile platform for sampling human antibody sequences under a variety of contexts relevant to antibody discovery and engineering. Model code and weights are publicly available at github.com/mgreenig/IgCraft.

Online Decomposition of Surface Electromyogram into Individual Motor Unit Activities Using Progressive FastICA Peel-off

Surface electromyogram (SEMG) decomposition provides a promising tool for decoding and understanding neural drive information non-invasively. In contrast to previous SEMG decomposition methods mainly developed in offline conditions, there are few studies on online SEMG decomposition. A novel method for online decomposition of SEMG data is presented using the progressive FastICA peel-off (PFP) algorithm. The online method consists of an offline prework stage and an online decomposition stage. More specifically, a series of separation vectors are first initialized by the originally offline version of the PFP algorithm from SEMG data recorded in advance. Then they are applied to online SEMG data to extract motor unit spike trains precisely. The performance of the proposed online SEMG decomposition method was evaluated by both simulation and experimental approaches. It achieved an online decomposition accuracy of 98.53% when processing simulated SEMG data. For decomposing experimental SEMG data, the proposed online method was able to extract an average of 12.00 +- 3.46 MUs per trial, with a matching rate of 90.38% compared with results from the expert-guided offline decomposition. Our study provides a valuable way of online decomposition of SEMG data with advanced applications in movement control and health.