Related Experiment Video
Updated: Jan 1, 2026
Author Spotlight: Advancing Early Detection and Treatment of Gastrointestinal Tumors
Published on: February 16, 2024
A Deep Learning Model for Cell Growth Inhibition IC50 Prediction and Its Application for Gastric Cancer Patients
Minjae Joo1, Aron Park1, Kyungdoc Kim2
1Department of Health Sciences and Technology, Gachon Advanced Institute for Health Sciences and Technology, Gachon University, Incheon 21999, Korea.
Abstract:
Heterogeneity in intratumoral cancers leads to discrepancies in drug responsiveness, due to diverse genomics profiles. Thus, prediction of drug responsiveness is critical in precision medicine. So far, in drug responsiveness prediction, drugs' molecular "fingerprints", along with mutation statuses, have not been considered. Here, we constructed a 1-dimensional convolution neural network model, DeepIC50, to predict three drug responsiveness classes, based on 27,756 features including mutation statuses and various drug molecular fingerprints. As a result, DeepIC50 showed better cell viability IC50 prediction accuracy in pan-cancer cell lines over two independent cancer cell line datasets. Gastric cancer (GC) is not only one of the lethal cancer types in East Asia, but also a heterogeneous cancer type. Currently approved targeted therapies in GC are only trastuzumab and ramucirumab. Responsive GC patients for the drugs are limited, and more drugs should be developed in GC. Due to the importance of GC, we applied DeepIC50 to a real GC patient dataset. Drug responsiveness prediction in the patient dataset by DeepIC50, when compared to the other models, were comparable to responsiveness observed in GC cell lines. DeepIC50 could possibly accurately predict drug responsiveness, to new compounds, in diverse cancer cell lines, in the drug discovery process.
Insights
DeepIC50, a novel deep learning model, predicts cancer drug responsiveness using genomic data and drug fingerprints. This approach improves accuracy for precision medicine and aids drug discovery in heterogeneous cancers like gastric cancer.
Area of Science:
- Computational biology
- Genomics
- Drug discovery
Background:
- Intratumoral cancer heterogeneity impacts drug response, necessitating precise prediction methods.
- Current drug responsiveness prediction models often overlook drug molecular features and patient mutation statuses.
- Gastric cancer (GC) is a lethal and heterogeneous malignancy with limited targeted therapy options.
Purpose of the Study:
- To develop and validate DeepIC50, a deep learning model for predicting cancer drug responsiveness.
- To incorporate mutation statuses and drug molecular fingerprints into drug responsiveness prediction.
- To evaluate DeepIC50's performance on pan-cancer cell lines and a gastric cancer patient dataset.
Main Methods:
- Construction of a 1-dimensional convolution neural network (DeepIC50).
- Inclusion of 27,756 features, encompassing mutation statuses and drug molecular fingerprints.
- Validation on two independent cancer cell line datasets and a real-world gastric cancer patient cohort.
Main Results:
- DeepIC50 demonstrated superior prediction accuracy for cell viability IC50 compared to existing models across pan-cancer cell lines.
- The model's predictions on the gastric cancer patient dataset showed comparable accuracy to cell line results.
- DeepIC50 effectively predicted drug responsiveness, highlighting its potential in precision oncology.
Conclusions:
- DeepIC50 offers a robust framework for predicting drug responsiveness in diverse cancer types.
- The model's ability to integrate genomic and drug features enhances precision medicine applications.
- DeepIC50 shows promise for accelerating drug discovery and development by predicting responses to novel compounds.
