ABSTRACT: Ion channel dysfunction underlies many diseases (e.g., arrhythmias, epilepsy, cystic fibrosis), and uncharacterized channels may also contribute to pathology. Identifying such channels and their residues directly contacting the permeation pathway (i.e., ion permeation residues) is key to elucidating transport mechanisms and developing targeted therapies. We developed BLAPE and CLAPE frameworks using the protein language model ESM-2 and curated datasets for high-throughput ion channel identification and permeation residue prediction. Our models achieved competitive performance, with relative improvements of 78% in Matthews correlation coefficients (MCC) on the ICIdentification dataset and 59% on the ICPermeation dataset compared with existing methods. Analysis of amino acid composition revealed enrichment for hydrophobic residues among ion permeation sites. Case studies on four diverse ion channels highlighted that CLAPE generally performs better than existing predictors and remains applicable to proteins lacking experimental structures, while also complementing structure-based pipelines such as AlphaFold3. We further applied our models to UniRef50 to predict potential ion channels, and made these results publicly available through the ICFinder webserver (https://tianlab-tsinghua.cn/icfinder/), providing a ready-to-use resource for the research community. All source code is available at https://github.com/JueWangTHU/ICFinder. BLAPE and CLAPE achieve competitive performance in ion channel identification and permeation residue prediction. ICFinder provides a practical resource for discovering potential ion channels, guiding target identification, and supporting drug design.

For detail: https://doi.org/10.1186/s12859-026-06561-6