scBasset

scBasset [1] (Python class SCBASSET) posits a sequence-based method for representation learning of scATAC-seq data.

The advantages of ScBasset are:

• Sequence representations allow for TF motif discovery and other sequence-based analyses.

• scBasset is fast and scalable.

The limitations of scBasset include:

• scBasset cannot currently leverage unobserved data and thus cannot curently be used for tranfer learning tasks.

Overview

TODO

---

[1]

Yuan Han and David R. Kelley (2022), scBasset: sequence-based modeling of single-cell ATAC-seq using convolutional neural networks, Nature Methods.