Transfer learning for drug-target interaction prediction

dc.authoridAtakan, Ahmet/0000-0001-9660-9758
dc.authoridAtalay, Volkan/0000-0001-7850-0601
dc.authoridDogan, Tunca/0000-0002-1298-9763
dc.authoridRifaioglu, Ahmet Sureyya/0000-0001-6717-4767
dc.authoridAcar, Aybar C./0000-0001-5694-8675
dc.authoridDalkiran, Alperen/0000-0002-4243-7281
dc.contributor.authorDalkiran, Alperen
dc.contributor.authorAtakan, Ahmet
dc.contributor.authorRifaioglu, Ahmet S.
dc.contributor.authorMartin, Maria J.
dc.contributor.authorAtalay, Rengul Cetin
dc.contributor.authorAcar, Aybar C.
dc.contributor.authorDogan, Tunca
dc.date.accessioned2025-01-06T17:45:10Z
dc.date.available2025-01-06T17:45:10Z
dc.date.issued2023
dc.description31st annual conference on Intelligent Systems for Molecular Biology (ISMB)/22nd European Conference on Computational Biology (ECCB) (ISMB/ECCB) -- JUL 23-27, 2023 -- Lyon, FRANCE
dc.description.abstractMotivationUtilizing AI-driven approaches for drug-target interaction (DTI) prediction require large volumes of training data which are not available for the majority of target proteins. In this study, we investigate the use of deep transfer learning for the prediction of interactions between drug candidate compounds and understudied target proteins with scarce training data. The idea here is to first train a deep neural network classifier with a generalized source training dataset of large size and then to reuse this pre-trained neural network as an initial configuration for re-training/fine-tuning purposes with a small-sized specialized target training dataset. To explore this idea, we selected six protein families that have critical importance in biomedicine: kinases, G-protein-coupled receptors (GPCRs), ion channels, nuclear receptors, proteases, and transporters. In two independent experiments, the protein families of transporters and nuclear receptors were individually set as the target datasets, while the remaining five families were used as the source datasets. Several size-based target family training datasets were formed in a controlled manner to assess the benefit provided by the transfer learning approach.ResultsHere, we present a systematic evaluation of our approach by pre-training a feed-forward neural network with source training datasets and applying different modes of transfer learning from the pre-trained source network to a target dataset. The performance of deep transfer learning is evaluated and compared with that of training the same deep neural network from scratch. We found that when the training dataset contains fewer than 100 compounds, transfer learning outperforms the conventional strategy of training the system from scratch, suggesting that transfer learning is advantageous for predicting binders to under-studied targets.
dc.description.sponsorshipTUBITAK [121E208]
dc.description.sponsorshipThis work was supported by TUBITAK project number:121E208.
dc.identifier.doi10.1093/bioinformatics/btad234
dc.identifier.endpagei110
dc.identifier.issn1367-4803
dc.identifier.issn1367-4811
dc.identifier.pmid37387156
dc.identifier.scopus2-s2.0-85163622323
dc.identifier.scopusqualityQ1
dc.identifier.startpagei103
dc.identifier.urihttps://doi.org/10.1093/bioinformatics/btad234
dc.identifier.urihttps://hdl.handle.net/20.500.14669/3343
dc.identifier.volume39
dc.identifier.wosWOS:001027457000016
dc.identifier.wosqualityQ1
dc.indekslendigikaynakWeb of Science
dc.indekslendigikaynakScopus
dc.indekslendigikaynakPubMed
dc.language.isoen
dc.publisherOxford Univ Press
dc.relation.ispartofBioinformatics
dc.relation.publicationcategoryKonferans Öğesi - Uluslararası - Kurum Öğretim Elemanı
dc.rightsinfo:eu-repo/semantics/openAccess
dc.snmzKA_20241211
dc.titleTransfer learning for drug-target interaction prediction
dc.typeConference Object

Dosyalar