cBioPortal/clinical-data-normalization

Name: clinical-data-normalization

Owner: cBioPortal

Description: Tools and data relating to clinical data normalization efforts.

Created: 2017-08-16 15:00:52.0

Updated: 2017-09-04 01:45:45.0

Pushed: 2017-09-04 01:45:44.0

Homepage: null

Size: 2555

Language: Python

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

cBioPortal-new-study-assistant

Introduction

This tool is designed to aid in curating new study clinical data into cBioPortal. The tool reads clinical data from the new study and compares the data to existing studies in the cBioPortal database to find attributes which match those in existing studies. The ultimate goal of the tool is to aid the curator in normalizing the new study data relative to cBioPortal studies.

Required python packages

• NumPy
• SciPy (recent version required)
• pandas
• seaborn
• python-levenshtein: anaconda link
• Pylatex (required only for the '–output_pdf' option)

Running the script

The default mode of the script selects a random study from the portal and searches other studies on the portal for matching attributes. The current version of the script typically takes around 10 minutes to run and depends on internet access to download data from cBioPortal. Alternatively, one can also clone the datahub repository and run the tool on that local data.

Default example:

on new_study_assistant.py

A more practical mode of the tool is to test raw study data from a new study against the existing cBioPortal data. Currently the tool assumes that the raw study data only contains one line in the header. Example raw data from the acyc_mda_2015 study is provided on this repository for reference.

Example using acyc_mda_2015 raw data (this data is provided in the acyc_mda_2015 folder in this repository):

on new_study_assistant.py --study_to_drop='acyc_mda_2015' --new_study_path='./acyc_mda_2015/raw_data_clinical.txt' > similarity_output.txt

Options available

w_study_path PATH

• Path to raw study data file that you want to analyze. Currently the code assumes that the file only contains attribute names in the header. If the file contains a multi-line header the program will probably crash.

udy_to_drop STUDY_ID

• Excludes a study (specified by STUDY_ID) from the analysis, this is useful when analyzing a study already on cBioPortal

ecific_study STUDY_ID

• Use this option if you want to run the analysis on a specific study (specified by STUDY_ID) that is already in the cBioPortal

tput_pdf

• Activate this flag if you would like the report results printed to a pdf instead of in an HTML format (requires Pylatex)

tahub_path PATH

• Specify path to a local version of the datahub, if this path is not specified the script will download data via the API instead

Output

Similar attributes that are detected in the test study are printed to a report. The report can be output as either an HTML(default) or PDF file.

Full list of the files generated by the script:

• report.html (or report.pdf) - HTML (or PDF) file containing an output report containing the attribute matches and several figures.
• dendrogram.png - An image showing the dendrogram obtained for similarity detection based on attribute values.
• n_attribute_distribution.png - An image showing the number of attributes contained in the test study relative to all studies on the cBioPortal.
• n_common_attribute_distribution.png - An image showing the number of attributes in common with studies on the cBioPortal.
• n_unique_attribute_distribution.png - An image showing the number of unique attributes in the test study relative to all studies on the cBioPortal.