Bayesian Sparse Matrix Factorization for Multimodal Integration
To modify CoGAPS to a parallel framework for efficiency and to input the underlying estimates of transcript abundance, transcriptional variance, and sparsity characteristic of single-cell RNA-sequencing data.
Results & Resources
The Fertig lab, in collaboration with Loyal Goff’s lab, completed a new parallel framework for CoGAPS that enables whole genome analysis of large single-cell datasets, openly available on github and bioconductor. Their work on CoGAPS is further described in their preprint. Another preprint describes the application of CoGAPS and ProjectR (Goff lab) to rapidly explore shared latent spaces across datasets and cellular measurement types.