IMGTgeneDL 0.5.0

Python script to download data from IMGT/GENE-DB

IMGTgeneDL

0.5.0

Jamie Heather | CCR @ MGH | 2023

This script provides an alternative way to access TCR and IG genes stored in IMGT/GENE-DB. It was mostly built to download human and mouse TCR sequences, so that's what is most tested, but it's readily adaptable to other species and loci.

Usage

This script is tested on python >= 3.6.

Installation

pip install IMGTgeneDL

Download specific gene types

The primary way this script is intended to be used is to tell it the species, loci, and sequence types that you want to download. This will be downloaded and saved to a file named with the date the IMGT release used, and details of the combination of parameters searched for (unless overriden with the -o / --out_path flag).

Species

The script must be run on single species at a time, given via the -s / --species flag as a full genus species with a '+' symbol in place of the space. E.g.:

• -s Homo+sapiens
• -s Mus+musculus

Note that it doesn't seem that the IMGT URL interface will accept either genus or species alone, and it can be particular about formatting so maintaining proper case is advised. However it seems to download sub-species (e.g.\ searching for Mus+musculus will return all covered strains).

Alternatively, users can use the -n / --common_names flag, and provide a common name. This only works for species that are included in the provided species.tsv file, which maps a common name into a scientific one for all the species for which sufficient TCR gene data exists in IMGT at the time of writing. This can be specified like so:

... -n -s human ... ... -n -s SHEEP ...

Loci

This script is currently configured to download the four common TCR loci:

• A / TRA / alpha
• B / TRB / beta
• G / TRG / gamma
• D / TRD / delta

These must be provided to the script using the -L / --loci flag, giving it the desired loci as a string of characters, e.g. -L AB to just download alpha and beta sequences, or -L G to just download gamma. Alternatively -L TR will simply download all four chains' sequences (equivalent to -L ABGD).

Sequence types

This script is designed to help in the aid in the analysis of typical expressed repertoires, and thus is configured by default to download the relevant parts of the loci that end up involved in expressed transcripts. The download of each is achieved using specific flags:

• -l / --get_l: download leader sequences
• -v / --get_v: download V sequences
• -d / --get_d: download D sequences
• -j / --get_j: download J sequences
• -c / --get_c: download constant region sequences

Note that these can be combined, e.g. -vdj will just download the V, D, and J gene sequences. Alternatively users can apply the -r / --get_all_regions flag to just download all of these regions (equivalent to -lvdjc).

Examples

The following is the basic command to download all relevant human sequences for all chains:

IMGTgeneDL.py -s Homo+sapiens -L TR -r

While this is a command to just download delta chain J genes from mice:

IMGTgeneDL.py -n -s mouse -j -L D

Download whole database

If no locus and gene type flags are used, or if the -a / --get_all flag is used, then the script will just download the whole of GENE-DB - all species, all genes, all loci. By default this downloads the ungapped nucleotide file, with all pseudogenes, and saves this to a file named with the date and the IMGT release used. This can be changed using the following flags:

• -gap / --gapped: downloads the gapped FASTA instead of the ungapped
• -ifp / --in_frame_p: downloads the 'inframeP' FASTA instead of the 'allP'
• -o / --out_path: as above, sets the path to a specific file if you don't wish to use the automatic names or save in the same directory

Output modes

This script has the option to specify different types of mode, selected via the -m / --output_mode flag. Currently there are two options: simple and stitchr.

simple mode

Simple mode outputs a single date-/version-stamped FASTA file of the requested loci/regions for a given species, as specified by the command line flag options described above.

stitchr mode

Stitchr mode downloads and formats sequence data suitable for use with Stitchr, the tool for generation of complete coding nucleotide sequences from minimal V/J/CDR3 information.

This mode downloads all loci for all regions that are available for the requested species, and saves them in a folder named after the common name of that species. That directory will contain the following files:

• data-production-date.tsv
  • Contains information about the IMGT and script versions used to generate this data
• imgt-data.fasta
  • Contains all of the FASTA reads that were successfully downloaded for this species
• J-region-motifs.tsv
  • Contains automatically inferred CDR3 junction ending motifs and residues (using the process established in the autoDCR TCR assignation tool), for use in finding the ends of junctions in stitchr
• C-region-motifs.tsv
  • Contains automatically inferred in-frame constant region peptide sequences, for use in finding the correct frame of stitched sequences
• TR[A/B/G/D].fasta
  • FASTA files of the individual loci's genes

It is simply run by specifying the mode and the desired species, either using common or scientific names:

IMGTgeneDL.py -s human -m stitchr -n
IMGTgeneDL.py -s Homo+sapiens -m stitchr

The script will only output a FASTA file for a specific locus if there is sufficient information for stitchr to use, i.e. there must be at least one leader, variable, joining, and constant region sequence, which are not currently available for all sequences that are listed in the database. In order to be used by stitchr it also appends an additional field to the end of the IMGT-provided FASTA header after a '~' character, labeling it's sequence type (LEADER/VARIABLE/JOINING/CONSTANT), allowing for explicit type declaration in the face of potentially variable IMGT-provided fields.

However note that while this script will generate files if those conditions are met, these data may not be sufficient for functional stitchr operation without manually adding sequences (e.g. if no matching leader/variable regions are found). It's recommended that users take care when using these tools for species with relatively little banked data.

Also note that stitchr mode will filter out any FASTA reads containing ambiguous or non-DNA residues, which are present for some species. All TRDV genes are also included in the TRA.fasta files by default, as (at least in humans) all can be found rearranged with TRAJ genes.

Notes

Constant region sequence downloading

The architecture of the TCR loci differs a little between the genes and across species, which the IMGT nomenclature has specific terms to cope with. However the URL based searching (or FASTA filtration in stitchr mode) in this script requires provision exact exon names for each species. However this presents an issue for constant regions, as these vart.

By default, IMGTgeneDL assumes the following exon configurations for each locus:

• TRAC: EX1+EX2+EX3+EX4UTR
• TRBC: EX1+EX2+EX3+EX4
• TRGC: EX1+EX2+EX3
• TRDC: EX1+EX2+EX3+EX4UTR

However requesting these complete exon arrangements automatically from IMGT often fails to produce a complete pre-spliced sequence, even when the correctly annotated individual exons appear to be present in the database. As such IMGTgeneDL instead downloads each exon individually and concatenates the sequences into complete constant regions as per the relevant exon orders. (This explains why '?' characters appear in the FASTA headers for constant region sequences, as this script doesn't re-calculate the relevant fields for assembled sequences, e.g. start/stop position in germline references.)

Alternative exon arrangements, as are common in gamma chain constant regions, can be specified in the c-region-variant-configs.tsv file. Note that in order to distinguish between potentially valid complete sequences from the same gene/allele combination but using a different arrangement of exons, IMGTgeneDL will append the non-standard characters to the end of the allele after an underscore. E.g. for humans, TRGC2*05 will specify the default exon arrangement 'EX1+EX2+EX3', while TRGC2*05_TR will specify the sequence for 'EX1+EX2T+EX2R+EX2+EX3'. It is recommended that users intending to use stitchr for a locus in a species with such variant configurations use the -p / preferred_allele_path option in stitchr to specify precisely which exon configuration they need for that constant region, as neither script is capable of determining the appropriate one to use automatically.

Users should also take caution as sometimes IMGT gene annotations can err, resulting in sequences that will not combine together downstream (e.g. in stitchr). The most common error I've noticed is that sometimes the 3' terminal residue of the J gene which gets spliced onto the first codon of the constant region can appear in the constant region sequence inappropriately, leading to a potential frameshift. It is recommended that users inspect the C-region-motifs.tsv file after running this script, and pay attention to the final column: this field specifies an amino acid motif downstream of any in-frame stop codons in a constant region. Ordinarily, the only loci which should have a sequence here are TRAC and TRDC (those with a EX4UTR exon), and thus a TRBC or TRGC gene with an entry here can be suggestive of an annotation error.

The other default IMGT labels downloaded are:

• L-PART1+L-PART2 for leader sequences
• V-/D-/J-REGION for V/D/J genes

IMGT FASTA headers

The IMGT header FASTA fields (as reported in the output of GENE-DB) are:

The FASTA header contains 15 fields separated by '|':

1. IMGT/LIGM-DB accession number(s)
2. IMGT gene and allele name
3. species
4. IMGT allele functionality
5. exon(s), region name(s), or extracted label(s)
6. start and end positions in the IMGT/LIGM-DB accession number(s)
7. number of nucleotides in the IMGT/LIGM-DB accession number(s)
8. codon start, or 'NR' (not relevant) for non coding labels
9. +n: number of nucleotides (nt) added in 5' compared to the corresponding label extracted from IMGT/LIGM-DB
10. +n or -n: number of nucleotides (nt) added or removed in 3' compared to the corresponding label extracted from IMGT/LIGM-DB
11. +n, -n, and/or nS: number of added, deleted, and/or substituted nucleotides to correct sequencing errors, or 'not corrected' if non corrected sequencing errors
12. number of amino acids (AA): this field indicates that the sequence is in amino acids
13. number of characters in the sequence: nt (or AA)+IMGT gaps=total
14. partial (if it is)
15. reverse complementary (if it is)

Licensing notes

The code and documentation of this repo are provided under an MIT license. I am not affiliated with IMGT, and this tool is only shared to help the field to better make of use of IMGT data. As stated on their website and in their publications, IMGT's policy regarding sharing of their data is that:

"... IMGT® software and data are provided to the academic users and NPO's (Not for Profit Organization(s)) under the CC BY-NC-ND 4.0 license. Any other use of IMGT® material, from the private sector, needs a financial arrangement with CNRS."

Please TCR responsibly.