High-precision high-coverage functional inference from integrated data sources

doi:10.1186/1471-2105-9-119

BMC Bioinformatics

Volume 9

Viewing options:

Abstract
Full text
PDF (657KB)
Additional files

Associated material:

Related literature:

Articles citing this article
on BioMed Central
on Google Scholar
on ISI Web of Science
on PubMed Central
Other articles by authors
on Google Scholar

Linghu B
Snitkin ES
Holloway DT
Gustafson AM
Xia Y
DeLisi C

on PubMed

Linghu B
Snitkin ES
Holloway DT
Gustafson AM
Xia Y
DeLisi C
Related articles/pages
on Google

on Google Scholar

on PubMed

Tools:

Post to:

Research article

High-precision high-coverage functional inference from integrated data sources

Bolan Linghu¹^,2 , Evan S Snitkin¹^,2 , Dustin T Holloway² , Adam M Gustafson¹^,2 , Yu Xia¹^,2 and Charles DeLisi¹^,2

¹Bioinformatics Graduate Program, Boston University, Boston, MA, 02215, USA

²Center for Advanced Genomic Technology, Boston University, Boston, MA, 02215, USA

author email corresponding author email

BMC Bioinformatics 2008, 9:119doi:10.1186/1471-2105-9-119


Published:	25 February 2008

Additional files

Additional File 1:

Supplemental material. Supplemental material provides additional detail descriptions about the following. (1) Brief descriptions of the four machine learning classifiers. (2) training/classification of individual classifiers and classifier aggregation. (3) Compare decision rules based on FLNs constructed by the other classifiers in yeast. (4) Detailed descriptions about two more prediction examples. (5) Compare the four decision rules for functional annotation in E. coli using linear SVM integrated FLN. (6) Compare decision rules in annotation precision-recall analysis. (7) Compare MW decision rule and functional flow algorithm. (8) Test robustness of the framework by reducing data sources or annotation sources. (9) Perform random control experiments for each evaluated annotation method.

Format: PDF Size: 396KB Download file

This file can be viewed with: Adobe Acrobat Reader

Additional File 2:

Prediction examples. Predicted annotations of un-annotated proteins not covered by KEGG pathways are defined as novel predictions, though some might have annotations in other databases such as SGD or MIPs [58,59]. We list more examples to show that our novel predictions with high estimated precisions represent appropriate pathway assignments using annotations in MIPs as supporting references. ORF names of proteins, predicted KEGG pathways by MW decision rules, estimated precisions based on the curve fitting function in figure 9, and annotations from MIPs database as supporting references are listed.

Format: XLS Size: 22KB Download file

This file can be viewed with: Microsoft Excel Viewer

Additional File 3:

Function-annotation predictions in yeast. List of predictions. Column 1: ORF name; column 2: predicted pathway annotation in KEGG pathway ID; column 3: MW score; column 4: estimated annotation precision; total number of proteins: 5475; total number of predicted annotations: 27,319; up to 5 annotations are predicted for each protein.

Format: TXT Size: 1.1MB Download file