The script aims to look at fragmented gene annotations (FRAGS) within prokka annotations. The FRAGS represent two (or more) ORFs that are in close proximity and are annotated with homology to the same gene. In such cases, Prokka ads an _n suffix to the gene ID. For example, a splitted genX can then be found as genX_1 and genX_2 in the GFF. See here for a case:

* The script will inform you how many case there is in your annotation. * If you think the FRAGS is due to a sequencing error (frameshift due to short indel), using the --frags parameter will fix the FRAGS if genX_1 and genX_2 are not in the same frame. The gff and the fasta file will be modified. The gene are merged, an insertion of one or two N will be added in between the genes to fix the frameshift. * If you think the FRAGS is not due to a sequencing error, use the --pseudo parameter, the gff will be fix (gene merged) and the agat_pseudo attribute (the value is the position of the codon stop) will be added to the related features. * using --frags and --pseudo is similar to use only --frags, except when no frameshift is found for a detected FRAGS (both gene are in the same frame), the agat_pseudo attribute is also added to the related features.

How the tool detecte the FRAGS? * Search for cases where contiguous genes have the same name (e.g. lpxA_1 lpxA_2). * If so we look at the size of the protein of each of those genes (lpxA_1 AA=175 ; lpxA_2 AA=116), and compute the size when merged togeter (devoided of the overlap if any) => here 270 AA * Then we look at the size of the protein used to infer the name (lpxA_1 inferred from Q9PIM1 = 263 AA ; lpxA_2 inferred from P0A722 = 262 AA ) and compute the average length of the reference protein: here 262AA. We add 20% to the length to be sure to include border cases => 282AA. * Compare the length of the merged proteins (262 AA) against the reference protein length (282). If the the expected protein length (282 AA) is longer we have a FRAGS.

SYNOPSIS -gff infile.gff --fasta genome.fa --db prokka/prokka_bacteria_sprot.fa  -o outfolder --help


  • --gff

    Input genome GTF/GFF file. Mandatory.

  • -f, --fa or --fasta

    Input genome fasta file. Mandatory.

  • --db

    Input Uniprot fasta file used by prokka. Mandatory.

  • --frags

    Merge and fix detected FRAGS if not in the same frame

  • --pseudo

    Merge detected FRAGS and add the agat_pseudo attribute (value will be the location of the first stop codon met).

  • --hamap_size

    Some protein function are not infered by Uniprot but by Hamap. In such case the information is retrieved from the web. As hamap provide a family profile, the protein size if a range. "low" option will use the low value of the range, "middle" option will use the average of the range, "high" option will the the high value of the range. Default "high".

  • --ct, --codon or --table

    Codon table to use. [default 1]

  • --skip_hamap

    For test purpose it could be useful to skip hamap, because it requires fetching information through internet.

  • -o , --output or --out

    Output folder. Mandatory.

  • -v

    verbose mode. Default off.

  • -c or --config

    String - Input agat config file. By default AGAT takes as input agat_config.yaml file from the working directory if any, otherwise it takes the orignal agat_config.yaml shipped with AGAT. To get the agat_config.yaml locally type: "agat config --expose". The --config option gives you the possibility to use your own AGAT config file (located elsewhere or named differently).

  • -h or --help

    Display this helpful text.