Skip to content

Releases: bm2-lab/UniTCR

First release of UniTCR

29 Mar 01:28
40495b0
Compare
Choose a tag to compare

UniTCR is a novel low-resource-aware multi-modal representation learning framework for unifiably integration and joint analysis of T cell receptor (TCR) and its corresponding transcriptome, which is composed of a dual-modality contrastive learning module and a modality preservation module. This model can be easily adapted to four applications in the computational immunology, including:

Single modality analysis
Modality gap analysis
Epitope-TCR binding prediction
Cross-modality generation