Coupling Deep Imputation with Multitask Learning for Downstream Tasks on Omics Data

Abstract

Omics data such as RNA gene expression, methylation and micro RNA expression are valuable sources of information for various clinical predictive tasks. For example, predicting survival outcomes, response to drugs, cancer histology type and other patients related information is possible using not only clinical data but molecular data as well. Moreover, using these data sources together, for example in multitask learning, might boost the performance. However, in practice, there are many missing data points which leads to significantly lower patient numbers when analysing full cases, which in our setting refers to all modalities being present. In this paper we investigate how imputing data with missing values using deep learning coupled with multitask learning can help to reach state-of-the-art performance results using combined omics modalities - RNA, micro RNA and methylation. We propose a generalised deep imputation method to impute values where a patient has data for one modality missing. Interestingly, deep imputation by itself outperforms multitask learning in classification and regression tasks across most combinations of modalities. In contrast, when using all available modalities for survival prediction we observe that multitask learning by itself significantly outperforms deep imputation (adjusted p-value of 0.03). Thus, both approaches are complementary when optimising performance for downstream predictive tasks.