A Systematic Approach to Featurization for Cancer Drug Sensitivity Predictions with Deep Learning

Abstract

By combining various cancer cell line (CCL) drug screening panels, the size of the data has grown significantly to begin understanding how advances in deep learning can advance drug response predictions. In this paper we train >35,000 neural network models, sweeping over common featurization techniques. We found the RNA-seq to be highly redundant and informative even with subsets larger than 128 features. We found the inclusion of single nucleotide polymorphisms (SNPs) coded as count matrices improved model performance significantly, and no substantial difference in model performance with respect to molecular featurization between the common open source MOrdred descriptors and Dragon7 descriptors. Alongside this analysis, we outline data integration between CCL screening datasets and present evidence that new metrics and imbalanced data techniques, as well as advances in data standardization, need to be developed.

References in corpus (9)

• Adam: A Method for Stochastic Optimization
• ADADELTA: An Adaptive Learning Rate Method
• Machine learning prediction of cancer cell sensitivity to drugs based on genomic and chemical properties
• Towards Explainable Anticancer Compound Sensitivity Prediction via Multimodal Attention-based Convolutional Encoders
• Chemception: A Deep Neural Network with Minimal Chemistry Knowledge Matches the Performance of Expert-developed QSAR/QSPR Models
• What is the Effect of Importance Weighting in Deep Learning?
• Towards Gene Expression Convolutions using Gene Interaction Graphs
• DeepCCI: End-to-end Deep Learning for Chemical-Chemical Interaction Prediction
• Improved Sampling Techniques for Learning an Imbalanced Data Set

Cited by in corpus (2)

• A cross-study analysis of drug response prediction in cancer cell lines
• AgEBO-Tabular: Joint Neural Architecture and Hyperparameter Search with Autotuned Data-Parallel Training for Tabular Data