MarkerCapsule: Explainable Single Cell Typing using Capsule Networks
Ray, S.; Schonhuth, A.
Show abstract
Many single cell typing methods require manual annotation which casts problems with respect to resolution of (sub-)types, manpower resources and bias towards existing human knowledge. The integration of heterogeneous data and biologically meaningful interpretation of results are further current key challenges. We introduce MarkerCapsule, which leverages the landmark advantages of capsule networks achieved in their original applications in single cell typing. Thereby, the small amount of labeled data required and the naturally arising, biologically meaningful interpretation of cell types in terms of characteristic gene activity patterns are exemplary strengths, beyond outperforming the state of the art in terms of basic typing accuracy. MarkerCapsule is available at: https://github.com/sumantaray/MarkerCapsule.
Matching journals
The top 7 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Predicting cell types with supervised contrastive learning on cells and their types 97%
- Unsupervised generative and graph representation learning for modelling cell differentiation 95%
- Correspondence analysis for dimension reduction, batch integration, and visualization of single-cell RNA-seq data 95%
Similar papers in this journal
- FIRM: Flexible Integration of single-cell RNA-sequencing data for large-scale Multi-tissue cell atlas datasets 96%
- CosGeneGate Selects Multi-functional and Credible Biomarkers for Single-cell Analysis 96%
- SMNN: Batch Effect Correction for Single-cell RNA-seq data via Supervised Mutual Nearest Neighbor Detection 96%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.