IGS Metagenomics Pipeline
This is a stable initial release of code that performs (via a combination of internal logic, wrappers for the khmer (0.8) package, and wrappers for IDBA-UD) both single- and paired-end assembly, with or without the help of a local SGE grid. Not all permutations are represented in this repository yet, but all four versions (single/paired, grid/no_grid) will appear in the near future.
DISCLAIMER
Until I get proper license text in place, everyone has permission to download, read and run this code; modify it locally all you want, but permission is expressly denied to -- in whole or in part -- include, repackage, release, publish, email, share or in any other way distribute either the original source or your subsequent modified version. If this arrangement should be found to violate any local laws, then those laws will supersede the arrangement's stated constraints only to the extent that they conflict with the law, with all non-conflicting constraints to remain in place.
INSTALLATION
-
Clone this repository onto your local machine.
a) If you have no thread-enabled SGE grid accessible to you, use the "*.LOCAL" codesets to run assemblies locally (make sure you have at least 100GB of memory).
b) If you do have a thread-enabled SGE grid available, then do the following in the "paired_reads.SGE_GRID" and "unpaired_reads.SGE_GRID" subdirectories:
Edit 000_setup.pl:
- Change the value of the "$INSTALL_ROOT" variable to point to your local clone directory (the parent directory of "paired_reads.SGE_GRID/").
- If your system requires that threaded jobs be specified with the "-pe threads X" argument to qsub (or something similar: the label following "pe" can be set at the installation level, and is thus not consistent across different grids), you'll need to modify the $SGEpeArg variable to conform to your own system.
- Change the name of $SGEdefaultQueue as needed (defaults to "all.q").
- Change the name of $SGEthreadedQueue as needed (defaults to "threaded.q").
OPERATION
Create a working directory for your assembly (you can specify multiple samples to be assembled in the same working directory if you wish).
Copy 000_setup.pl from the appropriate codeset to this new working directory and run it. By default, it'll prompt you for sampleID, mate and FASTQ-location data for your input sample(s), as well as for an SGE project code to use when running grid jobs (if you're using a grid). If you prefer, you can specify all of this information (as well as a list of grid nodes to exclude from submission, if applicable and desired) in three tab-delimited text files; see your cloned repository for examples of these files whose dummy values indicate correct syntax.
Run 003_run_assemblies.sh and wait. Associated grid job names (if running on a grid) will begin with "mga" and will contain the name of the associated sample ID; check qstat to assess completion. The final FASTA product for each assembled sample should appear in the sample's subdirectory of the main assembly working directory when done.