Skip to content

README.md

Latest commit

 

History

History
116 lines (83 loc) · 3.63 KB

README.md

File metadata and controls

116 lines (83 loc) · 3.63 KB

MutClust: Efficient and Scalable Mutual Rank-Based Coexpression Clustering

MutClust is a Python tool for efficient and scalable mutual rank-based gene coexpression analyses. The clustering analysis is conducted using ClusterONE, as described in Wisecaver et al. 2017. MutClust is still under development.

Features

  • Mutual Rank Analysis: Compute mutual rank (MR) from Pearson correlations on your gene expression matrix.
  • ClusterONE Clustering: Identify gene coexpression clusters from filtered/weighted MR networks.
  • Fast: Multi-threaded, sparse matrix operations for speed on large datasets.

Installation

Recommended

Install MutClust:

conda env create -f environment.yml
conda activate mutclust

Alternative

Step 1: Make sure that ClusterONE is available from the command line:

conda install bioconda::clusterone

Step 2a: Install MutClust from PyPI:

pip install mutclust

Step 2b: Or clone the repository from GitHub:

git clone https://github.com/eporetsky/mutclust.git
cd mutclust
pip install .

Usage

1. Calculate Mutual Rank (MR)

mutclust mr -i expr.tsv -o results.mrs.tsv.gz --mr-threshold 100 --threads 4 [--log2]
Argument Short Description Default
--input -i Path to the RNA-seq dataset (.tsv/.tsv.gz) Required
--output -o Output file for mutual rank pairs Required
--mr-threshold -m MR threshold for reporting gene pairs 100
--threads -t Number of CPU threads (correlation) 4
--log2 If set, applies log2(x+1) before calculation OFF by default
  • Input: Genes as rows, samples as columns (TSV, row index 'geneID').
  • Output: Gzipped tab-separated file containing Gene1, Gene2, MR.

2. Cluster Genes (with ClusterONE)

mutclust cls -i results.mrs.tsv.gz -o results.cls.tsv --e_value 10
Argument Short Description Default
--input -i Path to Mutual Rank (MR) pairs (.tsv/.tsv.gz) Required
--output -o Output file for clusters (.tsv) Required
--e_value -e Exponential decay constant for edge weighting 10
  • The tool filters/weights MR pairs and calls ClusterONE for clustering.
  • Output: clusters.tsv, listing clusters with p-value < 0.1. Tab-separated file containing clusterID, geneID, pval.

Example Workflow

mutclust mr -i data/myexpr.tsv -o out.mrs.tsv.gz --mr-threshold 100 --threads 72 --log2
mutclust cls -i out.mrs.tsv.gz -o out.clusters.tsv --e_value 10

Input Format

Expression file:

geneID\tSample1\tSample2\n...
GeneA \t1.1    \t2.2
GeneB \t4.2    \t3.7

Note: MutClust might be limited to linux because of dependency on pynetcor.

Coming Soon

  • Generate cluster gene annotation
  • Calculate cluster GO term enrichment
  • Calculate clusteer eigen-gene data
  • Add a MutClust Dockerfile
  • Add unit testing

License

MIT License. See LICENSE file for details.

Contributing

Suggestions, pull requests, and issues welcome!