mouthwash DSC

Dynamic Statistical Comparisons for MOUTHWASH simulations.

Setup

The necessary computational environment must be installed on a linux-64 machine. Below are instructions for using Docker or conda.

The Docker image is based on continuumio/miniconda3.

# Build image
docker build -t dsc-mouthwash .
# Run container
docker run --rm -it dsc-mouthwash
# Run Jupyter notebook
docker run -i -t -p 8888:8888 continuumio/miniconda3 /bin/bash -c "/opt/conda/bin/conda install jupyter -y --quiet && mkdir /opt/notebooks && /opt/conda/bin/jupyter notebook --notebook-dir=/opt/notebooks --ip='0.0.0.0' --port=8888 --no-browser --allow-root"

Alternatively if you have conda installed, you can install the environment directly on your machine.

conda create -n dsc-mouthwash \
  -c conda-forge -c bioconda -c defaults -c jdblischak \
  --file requirements/misc \
  --file requirements/python \
  --file requirements/r
source activate dsc-mouthwash

Notes

From Section 4.1:

simulate

• $p$ (genes) - 1000
• $\pi_0$ (proportion of null genes) - 0.5, 0.9, 1
• $n$ (samples) - 6, 10, 20, 40
• $m$ (control genes) - 10, 100

Number of combinbations = 3 * 4 * 2 = 24

500 replicates of each combination, therefore 500 * 24 = 12,000 simulated datasets.

analyze

1. OLS
2. SVA
3. CATErr
4. CATErr+MAD
5. RUV2
6. RUV3
7. CATEnc
8. CATEnc+MAD
9. CATEnc+Cal
10. MOUTHWASH
11. BACKWASH (our current plan is to skip this b/c it runs too long w/o sufficient improvement)

score

• AUC - ability to distinguish null versus non-null genes
• $\pi_0$ - estimated proportion of null genes

From section 4.3:

simulate

• GTEx tissue (23)
• List of control genes (2)

Therefore 23 * 2 = 46 datasets.

analyze

Same as above

score

• $s$ - number of sex-chromosome genes in top 100 most significant results
• $\pi_0$ - estimated proportion of null genes