Siamese MLP Improves Operon Prediction Over DGEB Baseline
A new computational method for operon identification, Siamese Contrastive Operon Pair Embeddings (SCOPE), outperforms the DGEB benchmark by using a Siamese MLP classifier over fused embedding spaces. Operon prediction is critical for understanding prokaryotic gene regulation, enabling regulatory network reconstruction and drug development. While experimental methods like RT-PCR and RNA-seq are accurate, they are labor-intensive and limited to model organisms, necessitating scalable computational approaches. Prior methods used logistic regression and decision trees as baselines. DGEB embeds sequences independently with a pre-trained protein language model and computes pairwise cosine similarity. SCOPE instead learns a classifier on fused embeddings, improving classification performance. The study was published on arXiv (2605.11022) as a cross-type announcement.
Key facts
- SCOPE uses a Siamese MLP classifier over fused embedding spaces
- DGEB benchmark uses independent embeddings with pairwise cosine similarity
- Operon identification is fundamental for understanding prokaryotic gene regulation
- Experimental methods like RT-PCR and RNA-seq are precise but laborious
- Prior computational approaches used logistic regression and decision trees
- The method was published on arXiv with ID 2605.11022
- The announcement type is cross
- SCOPE improves upon the DGEB baseline
Entities
Institutions
- arXiv