Transcriptome Deconvolution of Heterogeneous Tumor Samples with Immune Infiltration

Zeya Wang; Shaolong Cao; Jeffrey S. Morris; Jaeil Ahn; Rongjie Liu; Svitlana Tyekucheva; Fan Gao; Bo Li; Wei Lu; Ximing Tang; Ignacio I. Wistuba; Michaela Bowden; Lorelei Mucci; Massimo Loda; Giovanni Parmigiani; Chris C. Holmes; Wenyi Wang

doi:10.1016/j.isci.2018.10.028

Transcriptome Deconvolution of Heterogeneous Tumor Samples with Immune Infiltration

Zeya Wang, Shaolong Cao, Jeffrey S. Morris, Jaeil Ahn, Rongjie Liu, Svitlana Tyekucheva, Fan Gao, Bo Li, Wei Lu, Ximing Tang, Ignacio I. Wistuba, Michaela Bowden, Lorelei Mucci, Massimo Loda, Giovanni Parmigiani, Chris C. Holmes, Wenyi Wang

Dr. Bing Zhang Department of Statistics

Research output: Contribution to journal › Article › peer-review

60 Scopus citations

Abstract

Transcriptome deconvolution in cancer and other heterogeneous tissues remains challenging. Available methods lack the ability to estimate both component-specific proportions and expression profiles for individual samples. We present DeMixT, a new tool to deconvolve high-dimensional data from mixtures of more than two components. DeMixT implements an iterated conditional mode algorithm and a novel gene-set-based component merging approach to improve accuracy. In a series of experimental validation studies and application to TCGA data, DeMixT showed high accuracy. Improved deconvolution is an important step toward linking tumor transcriptomic data with clinical outcomes. An R package, scripts, and data are available: https://github.com/wwylab/DeMixTallmaterials.

Original language	English
Pages (from-to)	451-460
Number of pages	10
Journal	iScience
Volume	9
DOIs	https://doi.org/10.1016/j.isci.2018.10.028
State	Published - Nov 30 2018

Bibliographical note

Publisher Copyright:
© 2018

Funding

Z.W. and W.W. are supported by the U.S. National Cancer Institute through grant numbers R01CA174206, R01 CA183793, and P30 CA016672. W.W. is supported by U.S. National Cancer Institute 2R01 CA158113. Z.W. and J.S.M. are supported by NIH grants R01 CA178744 and P30 CA016672 and NSF grant 1550088. J.S.M. is supported by U.S. NSF 1550088 and the MD Anderson Colorectal Cancer Moonshot. S.C. J.A. X.T. and I.I.W. are supported by NIH grant 1R01CA183793. X.T. and I.I.W. are supported by The University of Texas Lung Specialized Programs of Research Excellence grant P50CA70907. S.T. is supported by NIH grant R01 CA174206 and Prostate Cancer Foundation Challenge Award. G.P. is supported by NIH grants 5R01 CA174206-05 and 4P30CA006516-51. M.L. is supported by NIH grants RO1CA131945, R01CA187918, DoD PC130716, P50 CA90381 and the Prostate Cancer Foundation, United States. We thank Vesteinn Porsson, Ilya Shmulevich, David Gibbs, Liuqing Yang, and Hongtu Zhu for useful discussions and valuable suggestions. Z.W. developed and coded the algorithms in DeMixT, analyzed the data, and performed the validation studies. S.C. performed the application study and analyzed the data using DeMixT. J.A. proposed the assumptions of linearity and model distributions. F.G. and R.L. helped build the DeMixT R package. S.T. B.L. W.L. X.T. I.I.W. M.B, L.M. and M.L. contributed data/materials for the validation and application studies. Z.W. and W.W. wrote the manuscript. J.S.M. S.T. G.P. and C.C.H. contributed to the discussion of results and revision of the manuscript. W.W. supervised the whole study. All authors read and approved the final manuscript. The authors declare no competing interests. Z.W. and W.W. are supported by the U.S. National Cancer Institute through grant numbers R01CA174206 , R01 CA183793 , and P30 CA016672 . W.W. is supported by U.S. National Cancer Institute 2R01 CA158113 . Z.W. and J.S.M. are supported by NIH grants R01 CA178744 and P30 CA016672 and NSF grant 1550088 . J.S.M. is supported by U.S. NSF 1550088 and the MD Anderson Colorectal Cancer Moonshot. S.C., J.A., X.T., and I.I.W. are supported by NIH grant 1R01CA183793 . X.T. and I.I.W. are supported by The University of Texas Lung Specialized Programs of Research Excellence grant P50CA70907 . S.T. is supported by NIH grant R01 CA174206 and Prostate Cancer Foundation Challenge Award. G.P. is supported by NIH grants 5R01 CA174206-05 and 4P30CA006516-51 . M.L. is supported by NIH grants RO1CA131945 , R01CA187918 , DoD PC130716 , P50 CA90381 and the Prostate Cancer Foundation, United States .

Funders	Funder number
National Sleep Foundation
National Science Foundation Arctic Social Science Program	1550088
National Institutes of Health (NIH)	R01 CA178744
UK Medical Research Council, Engineering and Physical Sciences Research Council	MC_UP_A390_1107
U.S. Department of Defense	PC130716, P50 CA90381
Prostate Cancer Foundation	5R01 CA174206-05, RO1CA131945, 4P30CA006516-51, R01CA187918
University of Texas Lung Specialized Programs of Research Excellence	R01 CA174206, P50CA70907
MD Anderson Colorectal Cancer Moonshot	1R01CA183793
National Childhood Cancer Registry – National Cancer Institute	R01 CA183793, P30 CA016672, R01CA174206, 2R01 CA158113

Keywords

Cancer
Computational Bioinformatics
Transcriptomics

ASJC Scopus subject areas

General

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1016/j.isci.2018.10.028

Cite this

@article{2b6ea3224637481c9ae3c38e0d434c3b,

title = "Transcriptome Deconvolution of Heterogeneous Tumor Samples with Immune Infiltration",

abstract = "Transcriptome deconvolution in cancer and other heterogeneous tissues remains challenging. Available methods lack the ability to estimate both component-specific proportions and expression profiles for individual samples. We present DeMixT, a new tool to deconvolve high-dimensional data from mixtures of more than two components. DeMixT implements an iterated conditional mode algorithm and a novel gene-set-based component merging approach to improve accuracy. In a series of experimental validation studies and application to TCGA data, DeMixT showed high accuracy. Improved deconvolution is an important step toward linking tumor transcriptomic data with clinical outcomes. An R package, scripts, and data are available: https://github.com/wwylab/DeMixTallmaterials.",

keywords = "Cancer, Computational Bioinformatics, Transcriptomics",

author = "Zeya Wang and Shaolong Cao and Morris, \{Jeffrey S.\} and Jaeil Ahn and Rongjie Liu and Svitlana Tyekucheva and Fan Gao and Bo Li and Wei Lu and Ximing Tang and Wistuba, \{Ignacio I.\} and Michaela Bowden and Lorelei Mucci and Massimo Loda and Giovanni Parmigiani and Holmes, \{Chris C.\} and Wenyi Wang",

note = "Publisher Copyright: {\textcopyright} 2018",

year = "2018",

month = nov,

day = "30",

doi = "10.1016/j.isci.2018.10.028",

language = "English",

volume = "9",

pages = "451--460",

}

TY - JOUR

T1 - Transcriptome Deconvolution of Heterogeneous Tumor Samples with Immune Infiltration

AU - Wang, Zeya

AU - Cao, Shaolong

AU - Morris, Jeffrey S.

AU - Ahn, Jaeil

AU - Liu, Rongjie

AU - Tyekucheva, Svitlana

AU - Gao, Fan

AU - Li, Bo

AU - Lu, Wei

AU - Tang, Ximing

AU - Wistuba, Ignacio I.

AU - Bowden, Michaela

AU - Mucci, Lorelei

AU - Loda, Massimo

AU - Parmigiani, Giovanni

AU - Holmes, Chris C.

AU - Wang, Wenyi

PY - 2018/11/30

Y1 - 2018/11/30

N2 - Transcriptome deconvolution in cancer and other heterogeneous tissues remains challenging. Available methods lack the ability to estimate both component-specific proportions and expression profiles for individual samples. We present DeMixT, a new tool to deconvolve high-dimensional data from mixtures of more than two components. DeMixT implements an iterated conditional mode algorithm and a novel gene-set-based component merging approach to improve accuracy. In a series of experimental validation studies and application to TCGA data, DeMixT showed high accuracy. Improved deconvolution is an important step toward linking tumor transcriptomic data with clinical outcomes. An R package, scripts, and data are available: https://github.com/wwylab/DeMixTallmaterials.

AB - Transcriptome deconvolution in cancer and other heterogeneous tissues remains challenging. Available methods lack the ability to estimate both component-specific proportions and expression profiles for individual samples. We present DeMixT, a new tool to deconvolve high-dimensional data from mixtures of more than two components. DeMixT implements an iterated conditional mode algorithm and a novel gene-set-based component merging approach to improve accuracy. In a series of experimental validation studies and application to TCGA data, DeMixT showed high accuracy. Improved deconvolution is an important step toward linking tumor transcriptomic data with clinical outcomes. An R package, scripts, and data are available: https://github.com/wwylab/DeMixTallmaterials.

KW - Cancer

KW - Computational Bioinformatics

KW - Transcriptomics

UR - http://www.scopus.com/inward/record.url?scp=85065297851&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85065297851&partnerID=8YFLogxK

U2 - 10.1016/j.isci.2018.10.028

DO - 10.1016/j.isci.2018.10.028

M3 - Article

AN - SCOPUS:85065297851

SN - 2589-0042

VL - 9

SP - 451

EP - 460

JO - iScience

JF - iScience

ER -

Transcriptome Deconvolution of Heterogeneous Tumor Samples with Immune Infiltration

Abstract

Bibliographical note

Funding

Keywords

ASJC Scopus subject areas

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this