Mechanistic modeling with variational autoencoders for bimodal single-cell RNA sequencing data
I will motivate and present a machine learning approach for mechanistic modeling of transcriptional dynamics. Current variational autoencoder frameworks ignore causal relationships between measurements from bimodal experiments, and I will show how they can be adapted to explicitly model biophysical processes. The analytical approach I will outline provides a generalizable strategy for treating multimodal datasets generated by high-throughput, single-cell genomic assays.
This is joint work with Maria Carilli, Gennady Gorin, Yongin Choi and Tara Chari