Identifier mapping performance for integrating transcriptomics and proteomics experimental results

BMC Bioinformatics

Methodology Article

Alex Lisovich¹  Kevin K McDade¹  Uma R Chandran¹  Roger S Day²  Traci Litzi³  David Kirchner⁴  VS Kumar Kolli⁴  Thomas P Conrads⁵  Brian L Hood⁵  G Larry Maxwell⁶ 

[1] Department of Biomedical Informatics, University of Pittsburgh School of Medicine, 15261, Pittsburgh, PA, USA;Department of Biomedical Informatics, University of Pittsburgh School of Medicine, 15261, Pittsburgh, PA, USA;Department of Biostatistics, University of Pittsburgh, 15260, Pittsburgh, PA, USA;Division of Gynecologic Oncology, Walter Reed Army Medical Center, 20307, Washington, D.C, USA;Windber Research Institute, 620 Seventh Street, 15963, Windber, PA, USA;Women's Health Integrated Research Center at Inova Health System, Inova Fairfax Hospital Campus, 3300 Gallows Road, 22042, Falls Church, VA, USA;Women's Health Integrated Research Center at Inova Health System, Inova Fairfax Hospital Campus, 3300 Gallows Road, 22042, Falls Church, VA, USA;Division of Gynecologic Oncology, Walter Reed Army Medical Center, 20307, Washington, D.C, USA;

关键词: Application Programming Interface;    Spectral Count;    Mapping Resource;    Bioinformatics Resource;    Annotation Resource;   

DOI  :  10.1186/1471-2105-12-213

 received in 2011-01-10, accepted in 2011-05-27,  发布年份 2011

来源: Springer

【 摘 要 】

BackgroundStudies integrating transcriptomic data with proteomic data can illuminate the proteome more clearly than either separately. Integromic studies can deepen understanding of the dynamic complex regulatory relationship between the transcriptome and the proteome. Integrating these data dictates a reliable mapping between the identifier nomenclature resultant from the two high-throughput platforms. However, this kind of analysis is well known to be hampered by lack of standardization of identifier nomenclature among proteins, genes, and microarray probe sets. Therefore data integration may also play a role in critiquing the fallible gene identifications that both platforms emit.ResultsWe compared three freely available internet-based identifier mapping resources for mapping UniProt accessions (ACCs) to Affymetrix probesets identifications (IDs): DAVID, EnVision, and NetAffx. Liquid chromatography-tandem mass spectrometry analyses of 91 endometrial cancer and 7 noncancer samples generated 11,879 distinct ACCs. For each ACC, we compared the retrieval sets of probeset IDs from each mapping resource. We confirmed a high level of discrepancy among the mapping resources. On the same samples, mRNA expression was available. Therefore, to evaluate the quality of each ACC-to-probeset match, we calculated proteome-transcriptome correlations, and compared the resources presuming that better mapping of identifiers should generate a higher proportion of mapped pairs with strong inter-platform correlations. A mixture model for the correlations fitted well and supported regression analysis, providing a window into the performance of the mapping resources. The resources have added and dropped matches over two years, but their overall performance has not changed.ConclusionsThe methods presented here serve to achieve concrete context-specific insight, to support well-informed decisions in choosing an ID mapping strategy for "omic" data merging.

【 授权许可】

CC BY   
© Day et al; licensee BioMed Central Ltd. 2011