Tag: bioinformatics

A new parallel deep learning framework combining CNN and Transformer branches with thermodynamic regularization improves the prediction of mutation-induced protein ...

A new graph neural network method called GLASS filters unreliable splice alignments in long-read RNA sequencing data, improving the precision ...

A new machine learning method called PrePssmCas fuses protein language model embeddings with evolutionary PSSM features to classify CRISPR-Cas proteins ...

A new graph neural network study of spatial transcriptomics finds that attention mechanisms, not elaborate edge features, drive accurate identification ...

A systematic benchmark shows that widely used microbiome mediation methods produce inflated false discoveries under compositional effects, prompting a new ...

A new framework called LLM-H2G fuses large language model semantics with hypergraph contrastive learning to predict herb–disease associations, outperforming existing ...

A new deep learning framework called SCmapST uses a seeded matches strategy to map single-cell gene expression data onto spatial ...

A new bioinformatics study links PTPN6 overexpression and promoter hypomethylation to overall survival in bladder cancer, positioning the gene as ...

A new transformer-based framework called THIMDA fuses multi-modal microRNA and disease representations to predict disease associations with high accuracy.

Researchers have developed a hyperboloid-based kernel method that maps biological sequence representations into hyperbolic space to overcome the distortion caused ...