MergeGI provides a single command line to merge and select barcoded raw data from MGI sequencing runs into a set of FastQ files ready for subsequent bioinformatics analysis.
We provide MergeGI as a Python library available on Pypi. The standalone application is called mergegi and can be installed in an environment with Python>3.7 as follows:
pip install mergegi
There is no dependencies except for click package so that installation should be straightforward.
For developers:
git clone git@github.com:sequana/MergeGI.git
cd MergeGI
poetry install
poetry shell
We also provide a pipeline for demultiplexing MGI data in Snakemake, which can be installed as an extra package:
pip install mergegi[sequana]
It installs snakemake and sequana_pipetools and you have access to sequana_mergegi command.
For developers:
git clone git@github.com:sequana/MergeGI.git
cd MergeGI
poetry install --all-extras
poetry shell
MergeGI is available as an apptainer image within the https://damona.readthedocs.io project. For example, the version 0.1.0 is available here: https://sandbox.zenodo.org/record/1134857/files/mergegi_0.0.1.img (60Mb).
The main goal of MergeGI is to select and merge the FastQ files generated by a MGI sequencer into a list of FastQ files directly usable for subsequent bioinformatics analysis. Why do we need to do this preprocessing ?
First, MGI generates one FastQ file per barcode. You may not need all those barcodes yet the demultiplexing performs a systematic search of all barcodes. Consequently, you will end up with FastQ files corresponding to your barcode and a bunch of FastQ files that should be ignored. Given the information from your wetlab colleagues you should have the list of samples and their relevant barcodes.
Second, MGI technologies imposes that barcodes being processed in a specific manner meaning that a given sample may be split into several barcodse (files). Therefore we need a tool to merge such files. Again, the wetlab should provide the barcodes corresponding to a given sample. See image below for more explanation
Third, a MGI flowcell has several lanes. You may want to merge the lanes or not.
Those 3 steps should be managed seemlessly by our tool given a sample sheet and the output directory of the MGI runs.
The data structure expected by MergeGI is the expected output directoy of MGI runs:
OutputFq/Flowcell/L01
OutputFq/Flowcell/L02
Where L01/L02 stands for lane 1 and 2.
The software needs a sample sheet that describe the sample name, the associated barcode identifier, the project name (it will be used to create the new output directory), and the lane where is the sample/barcode pair. Here is an example:
samplename,barcode,project,lane
A, 1, projectA, 1
B, 20, projectA, 1
A, 1, projectA, 2
C, 20, projectB, 2
C, 30, projectB, 1
B, 30, projectA, 2
If you have pooled a sample on the four lanes, meaning it is the same barcode on each lane, you can use the * character to simplify the sample sheet:
samplename,barcode,project,lane
A, 1, ,projectA, *
B, 20, ,projectA, *
IMPORTANT NOTE: The header must be present. The header names are not important but columns must be sorted with the expected order: sample name, barcode, project name, lane.
Given the sample sheet, and the input directory (top level of the MGI runs), this command should create a new clean directory with the relevant FastQ files (here in merge_data directory):
mergegi --samplesheet samplesheet.csv --input-directory mgi_raw_data --output-directory merge_data If the data is paired, add --paired argument
mergegi --samplesheet samplesheet.csv --input-directory mgi_raw_data --output-directory merge_data --pairedBy default, lanes are merged. If this is not what you want you may disable this option:
mergegi --samplesheet samplesheet.csv --input-directory mgi_raw_data --output-directory merge_data --paired --no-merge| Version | Description |
|---|---|
| 0.2.2 | use poetry and add sequana_mergegi as extra dependencies |
| 0.1.0 | simplify CI and use pyproject |
| 0.0.0 | firs release |
