Dataset Information

DeepCNV: a deep learning approach for authenticating copy number variations.

ABSTRACT: Copy number variations (CNVs) are an important class of variations contributing to the pathogenesis of many disease phenotypes. Detecting CNVs from genomic data remains difficult, and the most currently applied methods suffer from an unacceptably high false positive rate. A common practice is to have human experts manually review original CNV calls for filtering false positives before further downstream analysis or experimental validation. Here, we propose DeepCNV, a deep learning-based tool, intended to replace human experts when validating CNV calls, focusing on the calls made by one of the most accurate CNV callers, PennCNV. The sophistication of the deep neural network algorithm is enriched with over 10 000 expert-scored samples that are split into training and testing sets. Variant confidence, especially for CNVs, is a main roadblock impeding the progress of linking CNVs with the disease. We show that DeepCNV adds to the confidence of the CNV calls with an optimal area under the receiver operating characteristic curve of 0.909, exceeding other machine learning methods. The superiority of DeepCNV was also benchmarked and confirmed using an experimental wet-lab validation dataset. We conclude that the improvement obtained by DeepCNV results in significantly fewer false positive results and failures to replicate the CNV association results.

SUBMITTER: Glessner JT

PROVIDER: S-EPMC8681111 | biostudies-literature | 2021 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

DeepCNV: a deep learning approach for authenticating copy number variations.

Glessner Joseph T JT Hou Xiurui X Zhong Cheng C Zhang Jie J Khan Munir M Brand Fabian F Krawitz Peter P Sleiman Patrick M A PMA Hakonarson Hakon H Wei Zhi Z

Briefings in bioinformatics 20210901 5

Copy number variations (CNVs) are an important class of variations contributing to the pathogenesis of many disease phenotypes. Detecting CNVs from genomic data remains difficult, and the most currently applied methods suffer from an unacceptably high false positive rate. A common practice is to have human experts manually review original CNV calls for filtering false positives before further downstream analysis or experimental validation. Here, we propose DeepCNV, a deep learning-based tool, in ...[more]

PMID: 33429424

Similar Datasets

Project description:Pleomorphic xanthoastrocytoma (PXA) poses a diagnostic challenge. The present study relies on methylation-based predictions and focuses on copy number variations (CNV) in PXA. We identified 551 tumors from patients having received the histologic diagnosis or differential diagnosis pleomorphic xanthoastrocytoma (PXA) uploaded to the web page www.molecularneuropathology.org. Of these 551 tumors, 165 received the prediction "methylation class (anaplastic) pleomorphic xanthoastrocytoma" with a calibrated score >=0.9 by the brain tumor classifier version v12.8 and, therefore, were defined the PXA reference set designated mcPXAref. In addition to these 165 mcPXAref, 767 other tumors received the prediction mcPXA with a calibrated score >=0.9 but without a histological PXA diagnosis. The total number of individual tumors predicted by histology and/or by methylome based classification as PXA, mcPXA or both was 1318, and these were designated the study cohort. The selection of a control cohort was guided by methylation-based predictions recurrently observed for the other 386/551 tumors diagnosed as histologic PXA. 131/386 received predictions for another entity besides PXA with a score >=0.9. Control tumors corresponding to the 11 most common other predictions were selected, adding up to 1100 reference cases. CNV profiles were calculated from all methylation datasets of the study and control cohorts. Special attention was given to the 7/10 signature, gene amplifications and homozygous deletion of CDKN2A/B. Comparison of CNV in the subsets of the study cohort and the control cohort were used to establish relations independent of histological diagnoses. Tumors in mcPXA were highly homogenous in regard to CNV alterations, irrespective of the histological diagnoses. The 7/10 signature commonly present in glioblastoma, IDH-wildtype, was present in 15-20% of mcPXA, whereas amplification of oncogenes (likewise common in glioblastoma) was very rare in mcPXA (<1%). In contrast, the histology-based PXA group exhibited high variance in regard to methylation classes as well as to CNVs. Our data add to the notion, that histologically defined PXA likely only represent a subset of the biological disease.

Dataset Information

DeepCNV: a deep learning approach for authenticating copy number variations.

Publications

DeepCNV: a deep learning approach for authenticating copy number variations.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets