Abstract 2453: A new deconvolution algorithm for accurate assessing immune and stromal cell populations in mouse transcriptomic data

• Published in: Cancer research (Chicago, Ill.) Vol. 79; no. 13_Supplement; p. 2453
• Main Authors: Lu, Xiaoyu, Tu, Szu-wei, Chang, Wennan, Huo, Yan, Wang, Pengcheng, Zhang, Yu, Zhang, Chi, Cao, Sha
• Format: Journal Article
• Language: English
• Published: 01.07.2019
• ISSN: 0008-5472; 1538-7445
• DOI: 10.1158/1538-7445.AM2019-2453

Multiple deconvolution methods have been developed for investigating the heterogeneous immune and stromal (I/S) cell types in human cancer tissue by deciphering their relative abundances using transcriptomic data. However, there is a lack of a user-friendly software for mouse transcriptomic data deconvolution. Key challenges for developing such a tool include: (1) limited training data sets for available for deriving signature genes of I/S cell types in sporadic mouse data; (2) mouse models of diverse geno-/pheno-types may have varied expressions of I/S cell marker genes; (3) the transcriptomic data may be collected from mouse with certain levels of immuno-deficiency; and (4) the transcriptomic data may come from highly diverse experimental platforms. To solve these challenges, we (i) developed a novel non-parametric analysis method to derive potential I/S cell signature genes from a large collection of mouse data sets; (ii) implemented a low rank sub matrix identification method with a non-negative matrix factorization (NMF) based deconvolution method; and (iii) enabled the flexibility that certain I/S cell types may be absent if their respective cell markers do not form a significant low rank structure. The new method is applied to mouse prostate cancer data sets to infer the level of anti-cancer immune cell populations. Genes expressed by cancer cells that are negatively associated with the anti-cancer immune cells are further inferred, and compared with the results derived in human data. A user-friendly R package of the deconvolution method is released through GitHub. Citation Format: Xiaoyu Lu, Szu-wei Tu, Wennan Chang, Yan Huo, Pengcheng Wang, Yu Zhang, Chi Zhang, Sha Cao. A new deconvolution algorithm for accurate assessing immune and stromal cell populations in mouse transcriptomic data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2453.