RGASP3 coverage statistics pipeline

This repository contains the pipeline used for gathering and visualising feature coverage statistics for the RGASP3 project. The pipeline is built around a modified version of the HTSeq package.

The logic of the pipeline is contained in the analysis.mk file, please consult it to see how the tools below are chained together in order to perform the analyses.

Files

• bin - scripts build directory.
• config - files with tab separated fields specifying the style and content of the generated plots.
• lib/*.py - python classes used by the scripts under bin/
• plots - plots output directory.
• tools/HTSeq-0.5.3p3-rgasp3.tar.gz - the modified HTSeq python package used by the scripts.
• Makefile - makefile containing utility targets.
• analysis.mk - makefile containing analysis targets.
• sr_*.py - main tool source files.

Dependencies

• A UNIX environment with standard GNU tools and make
• python (>= 2.7.1)
• A modified version of the HTSeq package (download).
• numpy (>= 1.6.1)
• matplotlib (>= 1.1.0)
• The pipeline uses the Platform LSF workload manager to distribute the analysis between multiple compute nodes.

Building and using the tools

The following tools are being built under bin/ by issuing make:

annoparse

usage: annoparse [-h] -g gtf -l chromlens -p pickle_name [-s]
Parse and pickle annotation.
optional arguments:
-h, --help show this help message and exit
-g gtf Annotation in GFF format.
-l chromlens Chromosome list file (lengths ignored).
-p pickle_name Output pickle file.
-s Toggle stranded mode.

covstat

usage: covstat [-h] [-g annot_pickle] [-p pickle_prefix] input file
Harness feature coverage statistics.
positional arguments:
input file Input BAM file.
optional arguments:
-h, --help show this help message and exit
-g annot_pickle Pickled annotation.
-p pickle_prefix Output directory.

statvis

usage: statvis [-h] [-r report_pdf] -c color_file -m shape_file [-t title]
[-vs vs_file] [-vc cross_file] [-vp pc_file] [-xvs]
[input file [input file ...]]
Plot coverage statistics.
positional arguments:
input file Input pickled stats.
optional arguments:
-h, --help show this help message and exit
-r report_pdf Report PDF.
-c color_file Colors file.
-m shape_file Shapes file.
-t title Dataset title.
-vs vs_file Versus plots file.
-vc cross_file Cross plots file.
-vp pc_file Point correlation plots file.
-xvs Report list of valid stats.

statdump

usage: statdump [-h] [-g global_stats_file] [-t tr_stats_file]
[input file [input file ...]]
Dump primary alignment statistics.
positional arguments:
input file Input pickled stats.
optional arguments:
-h, --help show this help message and exit
-g global_stats_file Global stats file.
-t tr_stats_file Transcripts stats file.

Using the pipeline

After setting the relevant parameters in analysis.mk, the pipeline can be run by calling the following make targets:

• anno_parse - parse and pickle alignments.
• parse_sim - parse simulated BAM files.
• parse_mouse - parse mouse BAM files.
• parse_human - parse human BAM files.
• parse_human_stranded - parse human BAM files in stranded mode.
• plot_vs - plot selected coverage statistics for all datasets.
• plot_cross - produce cross-dataset plots.
• dump - dump primary alignment statistics to tab separated files.

Notes

• The logic for parsing the stranded paired-end reads (human datasets) is hard-coded in the parser class.