PBQDI Web Service

This web service provies Patient-Based Query and Data Integration (PBQDI) for the geWorkbench.

The input contains:

1. tumor type
2. the name of sample file
3. the content of the same file

The output contains:

1. tumor type
2. class assignments (map from sample names to subtype ID's)
3. result package as a zipped package

This zipped package includes the following entries:

• one HTML file
• one PDF file
• a number of image files. They are referenced from four different sections of the HTML file (the 'report'): one group for data quality, and the three groups for different categories of drugs, namely ontological, non-oncological, and investigational. The same image may show up in more than one categories of the drugs.

Dependency:

For this service to run, you need R installed, a number of required libraries, and the following R scripts and starting data files that are not included in the github repository due to copyright difference:

1. classifySamples.r
2. rununsupervised.r
3. properties.r
4. data-load-qc.r
5. norm-viper.r
6. oncoTarget-analysis.r
7. tumorSubtypes.txt
8. unsupervised.Rnw

dependency within R

• R version needs to be 3.3 or later
• 5 local packages need to be copied to the R library directory (specific location depending on your R installation). Their names are clinicalTrials, drugbank, n1database, n1platform, and tth. The actual code is not open-sourced.
  • Biobase: used by n1platform. It is not avaiable in the public repository. To install, see https://bioconductor.org/packages/release/bioc/html/Biobase.html (No everything works forthrightly from this site. I ended up fiddling around more with BioInstaller to get it work on Linux.)

Clients:

An example client program in Java is provided. The client does not store the ZIP file; instead, it unzips the contained files so they are ready to be used, e.g. the HTML file to be shown by a web browser or a browser component in other applications.