MultiMAP: dimensionality reduction and integration of multimodal data

• DOI: 10.1186/s13059-021-02565-y
• Publication Date: 2021-12-20T09:02:51Z
• Authors: Mika Sarkin Jain; Krzysztof Polanski; Cecilia Dominguez...; Xi Chen; Jongeun Park; Lira Mamanova; Andrew Knights; Rachel A. Botting; Emily Stephenson; Muzlifah Haniffa; Austen Lamacraft; Mirjana Efremova; Sarah A. Teichmann
• Keywords: Genetic Markers; 0301 basic medicine; QH301-705.5; Chromosome Mapping; Method; Genomics; QH426-470; Chromatin; 03 medical and health sciences; Gene Expression Regulation; Genetics; Chromosomes, Human; Humans; Biology (General); Single-Cell Analysis; Transcriptome; Algorithms; Software; Transcription Factors

Multimodal data is rapidly growing in many fields of science and engineering, including single-cell biology. We introduce MultiMAP, a novel algorithm for dimensionality reduction integration. MultiMAP can integrate any number datasets, leverages features not present all restricted to linear mapping, allows the user specify influence each dataset, extremely scalable large datasets. apply transcriptomics, chromatin accessibility, methylation, spatial show that it outperforms current approaches. On new thymus we use cells along temporal trajectory. This enables quantitative comparison transcription factor expression binding site accessibility over course T cell differentiation, revealing patterns versus opening kinetics.