Google

Spatial genomics foundation model that predicts a cell's transcriptome from its neighbours and forward-simulates tissue response to perturbation.

Assembly polishing model predicting base-level corrections to a diploid genome from PacBio HiFi read alignments, cutting base errors by about half.

Protein function prediction model that conditions a T5 encoder-decoder on retrieved homologs to assign EC numbers, GO terms and Pfam families.

Pfam-pretrained protein sequence models — transformer, LSTM and ResNet — released with five tasks benchmarking transfer from frozen embeddings.