Scholarlink

<p>Drug Target Interaction (DTI) prediction is vital for drug repurposing. Previous DTI studies on BioSNAP and BindingDB datasets often attribute biased predictions to “drug bias,” while our work reveals “target prior bias” as the predominant issue. This bias stems from the “prior tendency,” characterized by the imbalanced label distribution of targets in the training data. From causal lens, target “prior tendency” is a confounder, causing models trained with <i>P</i>(<i>Y</i>∣<b>D</b>,&#xa0;<b>T</b>) to learn spurious associations between targets and labels rather than genuine interaction mechanisms. In this study, we introduce alleviating <b>Ta</b>rget <b>P</b>rior <b>B</b>ias in Drug-Target Interaction Prediction (TAPB), a novel debiasing framework that employs amino acid randomization, confounder alignment module (CAM), and interventional training to compute <i>P</i>(<i>Y</i>∣<b>D</b>,&#xa0;<i>d</i><i>o</i>(<b>T</b>)) via backdoor adjustment, thereby addressing this bias. TAPB achieves competitive performance over existing approaches, demonstrating enhanced generalization and providing interpretable insights into DTIs.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

TAPB: an interventional debiasing framework for alleviating target prior bias in drug-target interaction prediction

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月2日

查看全文
  • Gaoming Lin,
  • Xin Zhang,
  • Zhonghao Ren,
  • Quan Zou,
  • Prayag Tiwari,
  • Changjun Zhou,
  • Yijie Ding

摘要

Drug Target Interaction (DTI) prediction is vital for drug repurposing. Previous DTI studies on BioSNAP and BindingDB datasets often attribute biased predictions to “drug bias,” while our work reveals “target prior bias” as the predominant issue. This bias stems from the “prior tendency,” characterized by the imbalanced label distribution of targets in the training data. From causal lens, target “prior tendency” is a confounder, causing models trained with P(Y∣D, T) to learn spurious associations between targets and labels rather than genuine interaction mechanisms. In this study, we introduce alleviating Target Prior Bias in Drug-Target Interaction Prediction (TAPB), a novel debiasing framework that employs amino acid randomization, confounder alignment module (CAM), and interventional training to compute P(Y∣D, do(T)) via backdoor adjustment, thereby addressing this bias. TAPB achieves competitive performance over existing approaches, demonstrating enhanced generalization and providing interpretable insights into DTIs.