EXTRACT_HOMOLOGS2

Identify target homologs in amino acid datasets

Disclaimer: This repository is maintained for the express purpose of retaining bioinformatic transparency with the code used for Tassia et al. 2017.

---

PREAMBLE ABOUT SCRIPT IN ITS ORIGINAL FORM:

The original function of this code was to identify TLR-pathway homologs across 40+ transcriptomic/genomic datasets. The original pipeline as used for Tassia et al. 2017 can be found as Extract_homologues_LEGACY.sh.

Each protein dataset is blasted against the SwissProt database and sequences which best-hit to target proteins are pulled and labeled as a putative homolog (merely by primary sequence similarity). To support homology given primary sequence similarity, the program annotates each sequence with SMART and Pfam domains to support homology and outputs the data in several parsable formats (InterProScan5's output formats). Following domain annotation, the user may use user-defined domain architectures to identify complete, non-target, divergent, or fragments given knowledge/literature on known domain architectures per homolog.

---

ABOUT EXTRACT_HOMOLOGS2:

EXTRACT_HOMOLOGS2 is a reimplimentation of its predacessor which was use in Tassia et al. 2017. Unlike its predacessor, EXTRACT_HOMOLOGS2 has been written to be more user-friendly and (ideally) be more transportable between individual HPCs. EXTRACT_HOMOLOGS2 is a bash script that must be made executable in a unix environment using chmod +x

DEPENDENCIES:

• DIAMOND
• SELECT_CONTIGS.PL - Written by James D. White; available through this distribution
• CD-HIT
• INTERPROSCAN

Note: If your HPC requires individual modules to be loaded into the environment, the four programs above will need to be loaded before running EXTRACT_HOMOLOGS2.

---

COMMAND LINE ARGUMENTS:

• Mandatory:

  -d <str> Path to diamond database file. To make a diamond databse, use the following command: diamond makedb --in [fasta file] --db [name of databse being made]
  -n <str> Naming template (example can be found in this distribution)
  -T <str> List of target SwissProt protein homologs (example can be found in this distribution)

• Optional:

  -h       Print help information
  -k       Keep intermediate files (disk-space expensive)
  -o <str> Output directory name (creates the directory; default value is the new directory ./Extract_homologs_output)
  -s <str> Directory containing peptide sequence files to be searched (default value is pwd)
  -S <str> Path to select_contigs.pl
  -t <int> Number of threads to be used with DIAMOND and InterproScan (default 1)

---

USAGE:

NOTE: For ease of use, all requisite programs should be installed into $PATH before using EXTRACT_HOMOLOGS2.

REQUIRED: For the program to run properly, the following must be considered:

• All searchable datasets should be translated prior to running the pipeline (commonly, we use TransDecoder - but program preference is up to the user).
• A renaming template (-n) and list of target homologs (-T) must be provided (see examples in this distribution).
• The protein datasets must have the .fasta extension. IMPORTANT: All files in the working directory with the .fasta suffix will be processed by EXTRACT_HOMOLOGS2. Therefore, insure all .fasta files in the working directory are peptide files intended for processing.
• If provided, -s argument must terminate with /.

EXAMPLE COMMAND:

EXTRACT_HOMOLOGS2.sh -d Blastdb/SwissProt.dmnd -n Naming_template.txt -T Target_homologs.txt -s Datasets/ -o TEST_OUTPUT2 -t 8 -k`

PREAMBLE ON INTERPROSCAN: First and foremost, Interproscan does not impact EXTRACT_HOMOLOGS2's annotation for putative homologs. Thus, if interproscan output files are not present in the Putative_homologs/ output directory, Interproscan can be run on the Putative_homologs_from_all_taxa.fasta file after EXTRACT_HOMOLOGS2 has finished processing.

The primary cause of Interproscan erroring out appears to be analysis incompatability with the host HPC. Built into the EXTRACT_HOMOLOGS2's implementation of Interproscan are the Pfam and SMART applications (see line 293). To ensure EXTRACT_HOMOLOGS2 will fully complete its annotation of the identified putative homologs, I recommend running the following command to confirm command compatability before running EXTRACT_HOMOLOGS2:

interproscan.sh -appl SMART -appl Pfam -i [input fasta file]

If running this command results in an application call error (see below), try activating the Pfam and SMART application packages.

18/01/2019 15:09:50:456 Welcome to InterProScan-5.17-56.0
Invalid input specified for -appl/--applications parameter:
Analysis 1 does not exist or is deactivated.

If other errors occur when running the interproscan command above, please refer to the InterProScan Wiki and/or contact your HPC admin.

---

OUTPUT:

Every successful run of EXTRACT_HOMOLOGS2 (when -k flag isn't provided) will output the following files:

• Putative_homologs_from_all_taxa.fasta - Sequences of putative homologs consolidated from all searched datasets
• Putative_homologs_from_all_taxa.interpro.gff3 - InterproScan output of Putative_homologs_from_all_taxa.fasta annotation in gff3
• Putative_homologs_from_all_taxa.interpro.tsv - InterproScan output of Putative_homologs_from_all_taxa.fasta annotation in tsv
• Putative_homologs_from_all_taxa.interpro.xml - InterproScan output of Putative_homologs_from_all_taxa.fasta annotation in xml
• Putative_homologs_from_all_taxa.blastp_vs_swissprot.outfmt6 - Diamond blastp output of Putative_homologs_from_all_taxa.fasta against SwissProt
• Putative_homologs_from_all_taxa.besthit_vs_swissprot.outfmt6 - Diamond blastp output organized into best-hit only of Putative_homologs_from_all_taxa.fasta against SwissProt
• [Fasta file of homologs per taxon] - Sequences of putative homologs organized into a single fasta file per dataset

---

For any questions, suggestions, or intention of use, please contact me at [email protected]

Michael Tassia, Auburn University