Variational Autoencoder for Anti-Cancer Drug Response Prediction
This work addresses the costly and slow process of drug discovery and therapy tailoring for cancer, though it is incremental as it builds on existing VAE and MLP methods.
The paper tackled predicting anti-cancer drug responses using variational autoencoders and multi-layer perceptrons, achieving high average R² scores of 0.83 for breast cancer and 0.845 for pan-cancer cell lines, and demonstrated the ability to generate new effective drug compounds.
Cancer is a primary cause of human death, but discovering drugs and tailoring cancer therapies are expensive and time-consuming. We seek to facilitate the discovery of new drugs and treatment strategies for cancer using variational autoencoders (VAEs) and multi-layer perceptrons (MLPs) to predict anti-cancer drug responses. Our model takes as input gene expression data of cancer cell lines and anti-cancer drug molecular data and encodes these data with our {\sc {GeneVae}} model, which is an ordinary VAE model, and a rectified junction tree variational autoencoder ({\sc JTVae}) model, respectively. A multi-layer perceptron processes these encoded features to produce a final prediction. Our tests show our system attains a high average coefficient of determination ($R^{2} = 0.83$) in predicting drug responses for breast cancer cell lines and an average $R^{2} = 0.845$ for pan-cancer cell lines. Additionally, we show that our model can generates effective drug compounds not previously used for specific cancer cell lines.