PARE: A framework for removal of confounding effects from any distance-based dimension reduction method

Dimension reduction tools preserving similarity and graph structure such as t-SNE and UMAP can capture complex biological patterns in high-dimensional data. However, these tools typically are not designed to separate effects of interest from unwanted effects due to confounders. We introduce the part...

Full description

Saved in:

Bibliographic Details
Published in:	PLoS computational biology Vol. 20; no. 7; p. e1012241
Main Authors:	Chen, Andrew A, Clark, Kelly, Dewey, Blake E, DuVal, Anna, Pellegrini, Nicole, Nair, Govind, Jalkh, Youmna, Khalil, Samar, Zurawski, Jon, Calabresi, Peter A, Reich, Daniel S, Bakshi, Rohit, Shou, Haochang, Shinohara, Russell T
Format:	Journal Article
Language:	English
Published:	United States Public Library of Science 10-07-2024 Public Library of Science (PLoS)
Subjects:	Algorithms Biology and Life Sciences Computational Biology - methods Computer and Information Sciences Ecology and Environmental Sciences Electronic data processing Genomics - methods Genomics - statistics & numerical data Graph theory Humans Medical colleges Medicine and Health Sciences Methods Neuroimaging - methods Research and Analysis Methods RNA sequencing Single-Cell Analysis - methods Single-Cell Analysis - statistics & numerical data Visualization (Computers) United States
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Dimension reduction tools preserving similarity and graph structure such as t-SNE and UMAP can capture complex biological patterns in high-dimensional data. However, these tools typically are not designed to separate effects of interest from unwanted effects due to confounders. We introduce the partial embedding (PARE) framework, which enables removal of confounders from any distance-based dimension reduction method. We then develop partial t-SNE and partial UMAP and apply these methods to genomic and neuroimaging data. For lower-dimensional visualization, our results show that the PARE framework can remove batch effects in single-cell sequencing data as well as separate clinical and technical variability in neuroimaging measures. We demonstrate that the PARE framework extends dimension reduction methods to highlight biological patterns of interest while effectively removing confounding effects.
Bibliography:	new_version ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 RB has received consulting fees from Bristol-Myers Squibb and EMD Serono and research support from Bristol-Myers Squibb, EMD Serono, and Novartis. DSR has received research funding from Abata Therapeutics, Sanofi-Genzyme, and Vertex Pharmaceuticals, all unrelated to the current study. RTS receives consulting income from Octave Bioscience and compensation for scientific reviewing from the American Medical Association.
ISSN:	1553-7358 1553-734X 1553-7358
DOI:	10.1371/journal.pcbi.1012241