Classification-based comparison of pre-processing methods for interpretation of mass spectrometry generated clinical datasets

doi:10.1186/1477-5956-7-19

Proteome Science

Volume 7

Viewing options:

Abstract
Full text
PDF (518KB)
Additional files

Associated material:

Related literature:

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

Wegdam W
Moerland PD
Buist MR
van Themaat EV
Bleijlevens B
Hoefsloot HC
de Koster CG
Aerts JM

on PubMed

Wegdam W
Moerland PD
Buist MR
van Themaat EV
Bleijlevens B
Hoefsloot HC
de Koster CG
Aerts JM
Related articles/pages
on Google

on Google Scholar

on PubMed

Tools:

Post to:

Research

Classification-based comparison of pre-processing methods for interpretation of mass spectrometry generated clinical datasets

Wouter Wegdam¹^* , Perry D Moerland²^* , Marrije R Buist¹ , Emiel Ver Loren van Themaat² , Boris Bleijlevens³ , Huub CJ Hoefsloot⁴ , Chris G de Koster³^,4 and Johannes MFG Aerts³

¹Department of Gynaecologic Oncology, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands

²Bioinformatics Laboratory, Department of Clinical Epidemiology, Biostatistics and Bioinformatics, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands

³Clinical Proteomics Group, Department of Medical Biochemistry, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands

⁴Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, the Netherlands

author email corresponding author email* Contributed equally

Proteome Science 2009, 7:19doi:10.1186/1477-5956-7-19


Published:	14 May 2009

Additional files

Additional file 1:

Variance analysis (ovarian cancer dataset). Boxplots of the coefficient of variation (CV, standard deviation/mean peak intensity). Left panel: CV for all combinations of pre-processing method (Ciphergen: cyan, Cromwell: red) and peak selection setting (A, B, C) for the CM10 chip. Right panel: idem for Q10 chip.

Format: PDF Size: 25KB Download file

This file can be viewed with: Adobe Acrobat Reader

Additional file 2:

Variance analysis (Gaucher dataset). Boxplots of the coefficient of variation (CV, standard deviation/mean peak intensity). CV for all combinations of pre-processing method (Ciphergen: cyan, Cromwell: red) and peak selection setting (A, B, C).

Format: PDF Size: 18KB Download file

This file can be viewed with: Adobe Acrobat Reader

Additional file 3:

Cumulative plot of significance of detected peaks (ovarian cancer dataset. CM10). For each combination of pre-processing method and peak selection settings, the cumulative percentage of peaks with a p-value smaller than the value on the x-axis are shown. P-value of a peak is based on a t-test between the normalized intensities of the cancer and the control group.

Format: PDF Size: 31KB Download file

This file can be viewed with: Adobe Acrobat Reader

Additional file 4:

Cumulative plot of significance of detected peaks (ovarian cancer dataset, Q10). For each combination of pre-processing method and peak selection settings, the cumulative percentage of peaks with a p-value smaller than the value on the x-axis is shown. P-value of a peak is based on a t-test between the normalized intensities of the cancer and the control group.

Format: PDF Size: 31KB Download file

This file can be viewed with: Adobe Acrobat Reader

Additional file 5:

Cumulative plot of significance of detected peaks (Gaucher dataset). For each combination of pre-processing method and peak selection settings, the cumulative percentage of peaks with a p-value smaller than the value on the x-axis is shown. P-value of a peak is based on a t-test between the normalized intensities of the Gaucher and the control group.

Format: PDF Size: 31KB Download file

This file can be viewed with: Adobe Acrobat Reader

Additional file 6:

Comparison of classifiers and pre-processing methods (ovarian cancer dataset). Average classification accuracy on 1000 test sets (size of training sets: 14, size of test sets: 14) for each specific combination of pre-processing method and peak selection settings.

Format: XLS Size: 24KB Download file

This file can be viewed with: Microsoft Excel Viewer

Additional file 7:

Comparison of classifiers and pre-processing methods (Gaucher dataset). Average classification accuracy on 500 test sets (size of training sets: 27, size of test sets: 12) for each specific combination of pre-processing method and peak selection settings.

Format: XLS Size: 19KB Download file

This file can be viewed with: Microsoft Excel Viewer

Additional file 8:

Comparison of classifiers and pre-processing methods (ovarian cancer dataset). Each combination of chip type, pre-processing method and peak selection was ranked by its average classification accuracy on 1,000 test sets (size of training sets: 14, size of test sets: 14) for each classifier. The heatmap gives a colour coding of the ranks from 1 (highest accuracy, red) to 18 (lowest accuracy, light yellow). Columns of the heatmap are ranked by their average rank over all classifiers, with Ciphergen pre-processing using setting C and the combined CM10/Q10 data getting the highest rank. Classifiers are ordered by their average rank over all pre-processing combinations, with DLDA being the best ranked classifier.

Format: PDF Size: 76KB Download file

This file can be viewed with: Adobe Acrobat Reader

Additional file 9:

Comparison of classifiers and pre-processing methods (Gaucher dataset). Each combination of pre-processing method and peak selection was ranked by its average classification accuracy on 500 test sets (size of training sets: 27, size of test sets: 12) for each classifier. The heatmap gives a colour coding of the ranks from 1 (highest accuracy, red) to 6 (lowest accuracy, light yellow). Columns of the heatmap are ranked by their average rank over all classifiers, with Ciphergen pre-processing using setting A getting the highest rank. Classifiers are ordered by their average rank over all pre-processing combinations, with SVM being the best ranked classifier.

Format: PDF Size: 44KB Download file

This file can be viewed with: Adobe Acrobat Reader