DSPLMF

DSPLMF predicts cancer drug sensitivity by applying logistic matrix factorization to genomic and chemical data to estimate probabilities of drug response for cell line–drug pairs.

Key Features:

• Logistic Matrix Factorization: Applies logistic matrix factorization with a regularization scheme to estimate probabilities of drug sensitivity across cell line–drug pairs.
• Similarity Integration: Incorporates cell line similarity from gene expression, copy number alterations, and single-nucleotide mutations and drug similarity from chemical structure to inform predictions.
• Data Utilization: Trains and evaluates models using CCLE (Cancer Cell Line Encyclopedia) and GDSC (Genomics of Drug Sensitivity in Cancer) datasets.
• Predicted IC50 Values: Produces predicted IC50 values as quantitative measures of drug potency for cell line–drug combinations.
• Latent Factors: Derives latent vectors from the matrix factorization that capture underlying relationships among cell lines and drugs.
• Performance Evaluation: Benchmarks against state-of-the-art methods and reports higher accuracy and efficiency in predicting drug responses.

Scientific Applications:

• Personalized Drug Sensitivity Prediction: Predicting cell line responses to anticancer drugs to support selection of therapies based on genomic profiles.
• Cancer Subtype Identification: Identifying distinct cancer subtypes within cell line collections by analyzing latent vectors derived from the model.
• Drug-Pathway Associations: Exploring associations between drugs and biological pathways using predicted IC50 values to aid interpretation of mechanisms of action and potential resistance.

Methodology:

Logistic matrix factorization with a novel regularization; incorporation of cell line similarity from gene expression, copy number alterations, and single-nucleotide mutations; incorporation of drug similarity from chemical structure; training and evaluation on CCLE and GDSC to predict IC50 and derive latent vectors; benchmarking against state-of-the-art methods.