Combining multiple hypothesis testing and affinity propagation clustering leads to accurate, robust and sample size independent classification on gene expression data
1 Biomedical Informatics Unit, Biomedical Research Foundation of the Academy of Athens, Athens, Greece
2 Department of Informatics and Telecommunications, National & Kapodistrian Univ. of Athens, Athens, Greece
3 Pharmacology Department, Medical School, National & Kapodistrian Univ. of Athens, Athens, Greece
BMC Bioinformatics 2012, 13:270 doi:10.1186/1471-2105-13-270Published: 17 October 2012
Additional file 1:
This function implements in R-code, the mAP-KL’s functionality.
Format: R Size: 6KB Download file
Additional file 2:
In this file, we present the 5-CV classification results for all real microarray data, when using three different classifiers (SVM-linear, KNN, and RF).
Format: XLSX Size: 35KB Download file
Additional file 3:
In this file, we present the Hold-out validation results for all real microarray data, when using three different classifiers (SVM-linear, KNN, and RF).
Format: XLSX Size: 24KB Download file
Additional file 4:
In this file, we have cited the subsets of genes according to the mAP-KL method.
Format: XLSX Size: 14KB Download file
Additional file 5:
Contains the microarray data used in this experiment. For each disease, we provide the ‘class_labels.csv’, ‘train.csv’ and ‘test.csv’ files, which represent the analogy of samples as described in table 1. The intensity values are unprocessed.
Format: ZIP Size: 15.3MB Download file
Additional file 6:
In this file, we have cited the clustering setup parameters, the DEGs position per simulation dataset, as well as the DEGs identified per method.
Format: XLS Size: 100KB Download file
This file can be viewed with: Microsoft Excel Viewer
Additional file 7:
In this file, we present the classification results of (mAP-KL, eBayes, maxT, RF-MDA) in the first simulation setup, where the clustering identification was under investigation. We employed three classifiers (SVM-linear, KNN, and RF).
Format: XLSX Size: 31KB Download file
Additional file 8:
In this file, we present the classification results in the ‘choedata’ when using two different mAP-KL’s subsets, stemming from two different ranking approaches. We used the SVM-linear, KNN, and RF classifiers to assess their performance.
Format: XLSX Size: 12KB Download file
Additional file 9:
This file contains the relevant scripts and functions for generating the simulated data. The ‘clusterSim’ r-package is required.
Format: ZIP Size: 5KB Download file