NearestTemplatePrediction (v4)

Summary

This module performs class prediction using a predefined list of marker genes for multiple (≥2) classes [1,2]. Such marker genes are usually selected based on fold change, t-statistic, correlation coefficient, regression coefficient, etc. These numerical values indicate the relative importance of each gene in the list and often help improve accuracy of the class prediction. The module can take such values as “weight” for 2-class prediction (optional). As measures of significance of the prediction for each sample, the module computes nominal p-value, false discovery rate (FDR) [3], and Bonferroni-corrected p-value based on a random resampling-based test.

Graphic output, including a heatmap of the marker genes, is generated if the users’ graphic device supports “png”. Alternatively, users can run the HeatMapViewer module to generate a heatmap of the marker genes based on the .gct and .cls files output by the NearestTemplatePrediction module.

References

1. Hoshida Y, et al. Gene Expression in Fixed Tissues and Outcome in Hepatocellular Carcinoma. N Engl J Med 2008; 359(19):1995-2004.
2. Xu L, et al. Gene Expression Changes in an Animal Melanoma Model Correlate with Aggressiveness of Human Melanoma Metastases. Molecular Cancer Research 2008;6:760-769.
3. Reiner A, Yekutieli D, Benjamini Y. Identifying differentially expressed genes using false discovery rate controlling procedures. Bioinformatics 2003;19(3):368-75.

Parameters

* - required

Input Files

1. <input.exp.filename>
   Gene expression dataset in GCT format.
2. <input.features.filename>
   Format for input features file:
   Tab-delimited text file. The first row contains column heads. The first column should be based on the same annotation system as the first column of the input gene expression (.gct) file.

Output Files

1. <output name>_prediction_result.xls
   Prediction result
2. <output name>_features.xls
   List of marker genes
3. <output name>_heatmap.png
   Heatmap of marker genes
4. <output name>_FDR_sample_bar.png
   Predicted sample labels at FDR < 0.05
5. <output name>_FDR.png
   Plot of FDR
6. <output name>_heatmap_legend.png
   Color map for SD –3~+3
7. <output name>_sorted.dataset.gct
8. <output name>_predicted_(un)sorted.cls
9. <output name>_sample_info.txt

Example Data

Example files from Hoshida, 2008 [1] are available: Train_Liver.gct, Hoshida_Survival_signature.txt

Task Type:
Prediction

CPU Type:
any

Operating System:
any

Language:
R-2.15.3