AI
Predicting Transmembrane Protein Topology from 3D Structure
Publication: cs.AI updates on arXiv.org
Author: Sitong Chen, Xiaopeng Mao
Source published: 28/09/2026, 04:00:00 UTC
Briefing created: 28/09/2026, 04:15:40 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 28/09/2026, 04:03:24 UTC.
Briefings
A new study proposes using the SchNet graph neural network to predict transmembrane protein topology, trained on the same dataset as the DeepTMHMM model. The model utilizes all atom-level embeddings rather than just sequences or alpha-carbons, showing potential without pre-trained weights. The research was published on arXiv in September 2026.
Key points
- The study introduces a novel approach using the SchNet graph neural network to infer transmembrane protein topology.
- The model was trained on the same dataset used for the DeepTMHMM model with 5-fold cross-validation.
Context and limitations
The findings are preliminary as the paper is a preprint on arXiv and the model has not yet been validated on external datasets.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 28/09/2026, 04:15:40 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeigpnpmbe22duqk2yavvydndpey6kja7qnef4pd54caubp33zwmb4m