Mittmich / pylola

Python implementation of LOLA

pyLOLA

Python implementation of LOLA (https://doi.org/10.1093/bioinformatics/btv612).

Installation

You can install this repo via pip as follows:

    pip install git+https://github.com/Mittmich/pylola

Usage

pyLOLA can be used to determine enrichment of a query genomic region with a set of target genomic regions in the context of a region universe (see the original LOLA paper for more details).

Regions are supplied as pandas dataframes and all overlap calculations are performed using bioframe.

A typical workflow starts by first loading a query region as follows:

    query = pd.read_csv(
            "./tests/test_files/query1.bed", sep="\t", header=None
        )

Then, one or more target regions are loaded to calculate enrichment amongst them:

    target1 = pd.read_csv(
            "./tests/test_files/target1.bed", sep="\t", header=None
        )

    target2 = pd.read_csv(
            "./tests/test_files/target2.bed", sep="\t", header=None
        )
    
        .
        .
        .
    targetN = pd.read_csv(
            "./tests/test_files/targetN.bed", sep="\t", header=None
        )

The "target-database" is constructed by putting these regions into a python iterable, for example a list:

    targets = [target1, target2, ..., targetN]

Finally, an appropriate universe is loaded that contains all possible regions:

    universe = pd.read_csv(
            "./tests/test_files/universe.bed", sep="\t", header=None
        )

Enrichment is then calculated using the pylola.run_lola function:

    result = pylola.run_lola(
            query, target_list, universe
        )

pylola.run_lola returns a dataframe that contains the flattened contingency table as well as the odds-ratio and negative logarithm of the p-value of region a fisher's exact test:

a	b	c	d	odds-ratio	p_value_log
39596	69998	919	1409	0.8673024	0.0001901352
.	.	.	.	.	.
.	.	.	.	.	.
.	.	.	.	.	.

The mapping to the respective contingency table is as follows:

	present in target	absent in target
present in query	a	b
absent in query	c	d

The order of the rows in the table corresponds to the order in the target_list. Additionally, one can pass an iterable containing the names of the respective target-items using the names parameters. Then, an additional name column will be appended to the output.