findBiomarkers

searches for candidate biomarkers in RNA Sequence data

Author: Andrew E. Davidson [email protected] https://github.com/aedavids/findBiomarkers

Starting notebooks

bin/startNotesbooks.sh

Installation

set up conda env with required packages

$ conda create --name findBiomarkers --file requirements.txt

you may need to install tensorflow and jupyter notebooks manually as follows

$ conda activate findBiomarkes
(findBiomarkers) $ pip install tensorflow
(findBiomarkers) $ pip install -q git+https://github.com/tensorflow/docs

(findBiomarkers) $ conda install -c conda-forge notebook
(findBiomarkers) $ conda install -c conda-forge jupyter_contrib_nbextensions

Installation notes

• clone https://github.com/aedavids/findBiomarkers
• download the required data set
  • you can find a copy at s3://bme-230a.santacruzintegration.com/tcga_target_gtex.h5
  • or run Rob Curries' ingest notebook

Running Unit test

cd ~workSpace/UCSC/findBiomarkers
conda activate findBiomarkers
export PYTHONPATH="${PYTHONPATH}:`pwd`/src"
cd src/test
python -m unittest discover .

Table of Contents

• ./documentation/dataDictionary

Jupyter notebooks

• lungCancerClassifierExploration.ipynb

  • basic exploration to get an idea of how to train a classifier

• cancerExploration.ipynb

  • checks to see if data set is balanced. Use to suggest data sub sets to train with

• lungCancerClassifier.ipynb

  • binary classifier

• lungCancerClassifierEvaluation.ipynb

  • How well does model work

• TCGA_Target_GTexPrototype.ipynb

  • used to develop dataUtilities/TCGA_Target_GTex.py

• testTCGA_Target_GTex.ipynb

  • test dataUtilities/TCGA_Target_GTex.py

Depecated Tree house notebooks The use a data set the compins the TCGA-target-GTex data sets with tree house childhood cancer

lungCancerClassifierExploration-TreeHouse.ipynb lungCancerClassifier-TreeHouse.ipynb lungCancerClassifierEvaluation-Treehouse.ipynb