Robust ATAC-seq Peak Calling for Many Samples via Convex Optimization
Project description
ROCCO: [R]obust [O]pen [C]hromatin Detection via [C]onvex [O]ptimization
What
ROCCO is an algorithm for efficient identification of "consensus peaks" in multiple HTS data samples (namely, ATAC-seq), where read densities are consistently enriched across samples or particularly strong enrichment is observed in a nontrivial subset of samples.
Example Behavior
In the image below, ROCCO is run on a set of ten heterogeneous ATAC-seq samples (lymphoblast) from independent donors (ENCODE).
- ROCCO consensus peaks are shown in red, where all default parameters are used in the first track, and the parametric-sigmoid transform
--use_parsig
option is applied to generate the results in the second track. - MACS2 (pooled library) consensus peak regions are shown in blue.
- ENCODE cCREs are included as a rough reference of potentially active regions, but note that these regions are not specific to the data samples used in this analysis, nor are they derived from the same cell type or assay.
How
ROCCO models consensus peak calling as a constrained optimization problem with an upper-bound on the total proportion of the genome selected as open/accessible and a fragmentation penalty to promote spatially consistency in active regions and sparsity elsewhere.
Why
ROCCO offers several attractive features:
- Consideration of enrichment and spatial characteristics of open chromatin signals
- Scaling to large sample sizes with an asymptotic time complexity independent of sample size
- No required training data or a heuristically determined set of initial candidate peak regions
- No rigid thresholds on the minimum number/width of supporting samples/replicates
- Mathematically tractable model with worst-case bounds on runtime and performance
Paper/Citation
If using ROCCO in your research, please cite the original paper in Bioinformatics (DOI: btad725
)
Nolan H Hamilton, Terrence S Furey, ROCCO: a robust method for detection of open chromatin via convex optimization,
Bioinformatics, Volume 39, Issue 12, December 2023
Documentation
For additional details, usage, etc. please see ROCCO's documentation: https://nolan-h-hamilton.github.io/ROCCO/
Note that using the module-level functions directly may allow for greater flexibility in applications than using the command-line interface, which is limited in scope.
Installation
PyPI (pip
)
pip install rocco
Build from Source
You can also build from source directly if preferred.
-
Clone or download this repository
git clone https://github.com/nolan-h-hamilton/ROCCO.git cd ROCCO python setup.py sdist bdist_wheel pip install -e .
ROCCO utilizes the popular bioinformatics software Samtools and bedtools. If not available already, these system dependencies can be installed with standard MacOS or Linux/Unix package managers, e.g., brew install samtools
(Homebrew), sudo apt-get install samtools
(APT).
Input/Output
- Input: Samples' BAM alignments or BigWig tracks
- Output: BED file of consensus peak regions
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.