Predicting drug activity against cancer through genomic profiles and SMILES
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.This study introduces two deep neural network models to predict anticancer drug effectiveness using genetic and chemical data. The models accurately predict drug response (IC50), improving personalized cancer treatment strategies.
Area Of Science
- Computational Biology
- Genomics
- Pharmacology
Background
- Cancer is a leading cause of death, necessitating improved detection and treatment.
- Personalized medicine aims to tailor therapies based on individual variability.
- Predicting tumor sensitivity to anticancer drugs remains a significant challenge.
Purpose Of The Study
- To develop deep neural network models for predicting anticancer drug impact.
- To estimate the half-maximal inhibitory concentration (IC50) of drugs on tumors.
- To integrate biological and chemical data for enhanced drug response prediction.
Main Methods
- Utilized Recurrent Neural Networks (RNNs) and Convolutional Neural Networks (CNNs).
- Pre-trained autoencoders on high-dimensional tumor gene expression and mutation data.
- Integrated genetic profiles and drug compound features for prediction.
Main Results
- RSEM outperformed TPM for gene expression data in deep models.
- CNNs provided better insights into gene expression data.
- Achieved a mean squared error of 1.06 in IC50 prediction, surpassing previous state-of-the-art models.
Conclusions
- Deep representations effectively predict drug potency (IC50).
- Proposed models demonstrate significant potential in personalized cancer therapy.
- The study advances the prediction of drug response in cancer cell lines.
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
