TL-iGeneCombo: Transfer Learning for Individual Sample Specific Gene Combination Effect Prediction and Selection in Cancer Cells.

Transfer learning in TL-iGeneCombo enhances prediction accuracy of gene combination effects in cancer cells, increasing Spearman correlation by 29% in prostate cancer.

• Why it matters: Accurate prediction of gene interactions is vital for developing effective cancer therapies, but limited CDKO data hampers model performance for individual cancer cells.
• What they did: Researchers developed TL-iGeneCombo, a transfer learning method that leverages CDKO data from other cancer cells to improve predictions of synthetic lethality and gene effects, applying it to prostate cancer 22RV1 cells.
• The result: The approach improved SL classification accuracy by 19% and validated superior gene combination effect prediction, enabling better targeted gene pair design for cancer treatment.