set_perm

Joshua Schmidt

• Introduction.
• A worked example
• Basic usage

Introduction.

set_perm is a python package to facilitate gene set enrichment tests i.e. GO set enrichment. It is heavily inspired by the excellent GOWINDA tool, developed by Robert Kofler.

While achieving similar results as GOWINDA, namely feature (gene) lengh bias correction, PySetPerm adds additional functionality.

First, set_perm enables testing of depletion in addition to enrichment.

Second, set_perm enables testing of the enrichment/depletion of combined candidate sets (joint distribution). The idea is best illustrated by an hypothetical example.

Suppose a method to detect loci under positive selection is used in popA and popB, identifying several candidate loci in each population. We could test if candidates in either of the populations are enriched for particular functions. This would be a run of the mill GO test for popA and popB individually, where we might be lucky to find a significant enrichment in the same gene set for both taxa. But this is likely underpowered.

Perhaps a more interesting question, biologically and statistically, is to ask if the joint distribution of popA and popB candidate loci is enriched for a particular function - this could suggest that a common pathway has been under selection.

The aim for set_perm is to develop a simple, extensible framework for developing these tests. At the moment simple testing of joint candidate set distributions is implemented. Future plans include equivalent explicit union and intersect tests, varied block permutation options and more!

A worked example

Go to the notebook here for an example workflow: chimp example

Basic usage

set_perm \  
--candidates "central,data/central-0.000192-candidate.snps.bed.gz" \
             "eastern,data/eastern-0.000228-candidate.snps.bed.gz" \
             "ancestral,data/ancestral-0.005-candidate.snps.bed.gz" \
--background "central,data/pbsnj-bg.snps.bed.gz" \
             "eastern,data/pbsnj-bg.snps.bed.gz" \
             "ancestral,data/ancestral-bg.bed.gz" \
--feature_def data/genes.txt \
--function_def data/kegg.txt \
--min_set_size 10 \
--n_perms 10000 \
--prefix "test" \
--joint "central,eastern" "central,ancestral" \
        "eastern,ancestral" "central,eastern,ancestral" \
--gene_def 2000 \
--threads 10

Units of analysis

Create objects

join objects

.join_objects() functions for both Input and SetPerPerm objects enable fast creation of union sets.