Removal of batch effects using distribution-matching residual networks

• Published in: Bioinformatics (Oxford, England) Vol. 33; no. 16; pp. 2539 - 2546
• Main Authors: Shaham, Uri, Stanton, Kelly P, Zhao, Jun, Li, Huamin, Raddassi, Khadir, Montgomery, Ruth, Kluger, Yuval
• Format: Journal Article
• Language: English
• Published: England Oxford University Press 15.08.2017
• Subjects: Computational Biology - methods; Cytophotometry - methods; Data Accuracy; Humans; Machine Learning; Original Papers; Sequence Analysis, RNA - methods; Single-Cell Analysis - methods; Statistics as Topic
• ISSN: 1367-4803; 1367-4811; 1367-4811
• DOI: 10.1093/bioinformatics/btx196

Sources of variability in experimentally derived data include measurement error in addition to the physical phenomena of interest. This measurement error is a combination of systematic components, originating from the measuring instrument and random measurement errors. Several novel biological technologies, such as mass cytometry and single-cell RNA-seq (scRNA-seq), are plagued with systematic errors that may severely affect statistical analysis if the data are not properly calibrated. We propose a novel deep learning approach for removing systematic batch effects. Our method is based on a residual neural network, trained to minimize the Maximum Mean Discrepancy between the multivariate distributions of two replicates, measured in different batches. We apply our method to mass cytometry and scRNA-seq datasets, and demonstrate that it effectively attenuates batch effects. our codes and data are publicly available at https://github.com/ushaham/BatchEffectRemoval.git. yuval.kluger@yale.edu. Supplementary data are available at Bioinformatics online.