Deep learning for inferring gene relationships from single-cell expression data
Problem
Inferring gene-gene relationships (coexpression, functional assignment, pathway reconstruction, causality) from single-cell expression data.
Why it matters
Understanding gene regulation underpins disease-gene identification and pathway knowledge in biology and medicine.
What the AI did
The CNNC framework encodes gene expression data and applies deep (convolutional) neural networks to address diverse gene-relationship tasks in one framework.
Results
Claimed improvement over prior methods on transcription factor targets, disease-related genes and causality inference; concrete numbers not stated in source.
Impact
Provides a flexible framework extendable to additional genomics data types, per abstract.
Record
| Venue | Proceedings of the National Academy of Sciences |
|---|---|
| Published | 2019 |
| Work type | article |
| Authors | Ye Yuan, Ziv Bar‐Joseph |
| Citations | 265 (OpenAlex, at import time) |
| References | 51 |
| Model or method | CNNC (convolutional neural network for coexpression) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1073/pnas.1911536116 |
| Open access | green |
| Keywords | deep learning, CNNC, single-cell, coexpression, causality inference, disease genes |
Importance rating
real but narrower advance
Gene-relationship inference underpins disease-gene identification and pathway knowledge; broad relevance but abstract claims only relative improvement, no scale.