Software
antigen-forecasting
Analysis scripts and tools for antigen-prime simulations. Includes variant assignment benchmarking (k-means antigenic, pathogen-embed sequence-based, Neher phylogenetic), growth rate inference with evofr, and metrics like normalized information distance (NID) for comparing variant assignments.
antigen-prime
Forward-time epidemic simulator coupling genetic and antigenic viral evolution under host immunity selection. Extends Bedford’s antigen simulator by adding explicit nucleotide sequences with K80 mutation model. Useful for generating ground-truth data for benchmarking variant assignment and growth rate inference methods.
polyclonal
Models polyclonal antibody binding and escape using a biophysical framework. Decomposes serum into epitope-specific activities, infers per-mutation escape at each epitope, and provides tools for fitting models, predicting escape, and plotting epitope maps.
torchdms
PyTorch models for deep mutational scanning (DMS) functional score prediction. Implements global epistasis models including GGE (additive latent effects + nonlinear transform), independent latent spaces per phenotype, and conditional architectures where one phenotype depends on another.