Format

Send to

Choose Destination
iScience. 2020 Jan 24;23(1):100780. doi: 10.1016/j.isci.2019.100780. Epub 2019 Dec 18.

Scalable Prediction of Acute Myeloid Leukemia Using High-Dimensional Machine Learning and Blood Transcriptomics.

Author information

1
LIMES-Institute, Department for Genomics and Immunoregulation, University of Bonn, Carl-Troll-Str. 31, 53115 Bonn, Germany.
2
Statistics and Machine Learning, German Center for Neurodegenerative Diseases, Venusberg-Campus 1, Building 99, 53127 Bonn, Germany.
3
PRECISE Platform for Single Cell Genomics and Epigenomics, German Center for Neurodegenerative Diseases and the University of Bonn, Venusberg-Campus 1, Building 99, 53127 Bonn, Germany.
4
Molecular Immunology in Neurodegeneration, German Center for Neurodegenerative Diseases, Venusberg-Campus 1, Building 99, 53127 Bonn, Germany; PRECISE Platform for Single Cell Genomics and Epigenomics, German Center for Neurodegenerative Diseases and the University of Bonn, Venusberg-Campus 1, Building 99, 53127 Bonn, Germany.
5
MLL, Münchner Leukämielabor GmbH, Max-Lebsche-Platz 31, 81377 München, Germany.
6
Statistics and Machine Learning, German Center for Neurodegenerative Diseases, Venusberg-Campus 1, Building 99, 53127 Bonn, Germany. Electronic address: sach.mukherjee@dzne.de.
7
LIMES-Institute, Department for Genomics and Immunoregulation, University of Bonn, Carl-Troll-Str. 31, 53115 Bonn, Germany; PRECISE Platform for Single Cell Genomics and Epigenomics, German Center for Neurodegenerative Diseases and the University of Bonn, Venusberg-Campus 1, Building 99, 53127 Bonn, Germany. Electronic address: j.schultze@uni-bonn.de.

Abstract

Acute myeloid leukemia (AML) is a severe, mostly fatal hematopoietic malignancy. We were interested in whether transcriptomic-based machine learning could predict AML status without requiring expert input. Using 12,029 samples from 105 different studies, we present a large-scale study of machine learning-based prediction of AML in which we address key questions relating to the combination of machine learning and transcriptomics and their practical use. We find data-driven, high-dimensional approaches-in which multivariate signatures are learned directly from genome-wide data with no prior knowledge-to be accurate and robust. Importantly, these approaches are highly scalable with low marginal cost, essentially matching human expert annotation in a near-automated workflow. Our results support the notion that transcriptomics combined with machine learning could be used as part of an integrated -omics approach wherein risk prediction, differential diagnosis, and subclassification of AML are achieved by genomics while diagnosis could be assisted by transcriptomic-based machine learning.

KEYWORDS:

Artificial Intelligence; Biological Sciences; Cancer; Computer Science; Omics; Transcriptomics

Conflict of interest statement

Declaration of Interests There are no competing interests.

Supplemental Content

Full text links

Icon for Elsevier Science Icon for PubMed Central
Loading ...
Support Center