ref: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8221386/
Kmerator is a prototype tool designed for the prediction of specific k-mers (also called tags) from input sequences, considering a reference genome and an ENSEMBL-like transcriptome. From these specific k-mers, it also outputs their corresponding specific contigs which are sequences of consecutive k-mers (overlapping length between k-mers must be k-1, otherwise, it's a new contig). You need to provide kmerator with a jellifsh index of the reference genome. Kmerator itself builds a jellyfish index of the reference transcriptome (by default the latest available version of Ensembl). It then decomposes your input transcript or gene sequences to count the occurences of each k-mer in the genome and transcriptome. Number of occurrences are then interpreted, in different manners, to select specific k-mer from your input.
Before using kmerator, a jellyfish index of the reference genome must be created. kmerator automatically creates a dataset according to the species and the desired release number (by default, homo_sapiens and the latest version). The dataset is composed of 4 files per species/version: a jellyfish index of the modified transcriptome (cDNA + ncRNA - alternative chormosomes) from Ensembl, a binary file representing the same transcriptome, another binary file containing general information on the genes of the transcriptome and a report file.
- Python >= v3.7
- Jellyfish >= 2.0
Install with pip
pip3 install kmerator
With pip, it is also easy to install an older version
pip install kmerator==1.0.0
Installation from github
git clone https://github.com/Transipedia/kmerator.git
ln -s $PWD/kmerator3/kmerator/kmerator.py /usr/local/bin/kmerator # or somewhere in your $PATH
Before all, remember that kmerator needs a jellyfish index of the genome. You must build it according to the species you are studying. You can store and name the index file whatever you want. Please note that you must use the jellyfish -C option when building the reference genome index.
The arguments to run kmerator are numerous, so to reduce the number of arguments to enter, it is advisable to edit the configuration file with the command :
kmerator -e
By filling in the datadir and genome directives, you will avoid having to re-enter the --datadir and --genome arguments systematically. If you are working on a species other than Human, you can also fill in the specie directive. And in a long-term project, you may want to set a release number.
There are two main cases:
- you find for specific k-mers for annotated genes or transcripts : use the
--selectionoption, followed by:- the list of gene and/or transcripts separated by a space
- or a file with the list of genes/transcripts. Separator could by a space, a tab or a newline, and comments are allowed (
#)
- you find for specific k-mers of unannotated sequences : use the
--fasta-fileoption, followed by a fasta file containing yours requests. In case of you focuses on chimeras, add the--chimeraoption
Examples:
kmerator -s npm1 brca2 ENST00000255409 ENSG00000159216 # you can mix genes and transcripts
kmerator -s genes.txt # you can also use a file with gene list
kmerator -f file.fa # give a fasta file fr unannotated sequences
Note the above commands assume that the configuration file contains at least the datadir and genome directives, the default species is homo_sapiens and the last available version will be used (if it is not present in datadir, kmerator will propose the construction of a dataset automatically)
- When you are looking for specific kmers of a gene (symbol, alias or Ensembl name), kmerator fetch sequence of its canonical transcript, extracts kmers and keep those that found only in the gene.
- When you are looking for a transcript, kmerator only keeps the kmer found in the transcript, and only in that transcript. If isoforms completely cover the transcript, no kmer will be kept.
To work, kmerator needs a jellyfish index of the genome, a jellyfish index of the transcriptome and various files. You will have to make the jellyfish genome index manually. Instead, kmerator builds the jellyfish transcriptome index and the files it needs, which we call datasets. There is one dataset per species and per transcriptome version. When kemrator does not find (in datadir) the requested transcriptome release (by default, the latest available on Ensembl), it offers to automatically build the dataset in question. In addition, dataset management options are available:
kmerator -l # list local datasets
kmerator -u # find last release on Ensembl, and build dataset if not present
kmerator --mk-dataset # build dataset according to -r <release> and -S <specie> arguments
kmerator --rm-dataset # delete dataset according to -r <release> and -S <specie> arguments
You can get information about gene, using the --info parameters. Like previous argument, you can enter a mix of symbol gene, Ensembl gene or Ensembl transcript (ex: braf, ENSG00000157764, ENST00000646427), or even use a text file with the list of requested genes separated by space, tab or newline (comments: '#').
In addition, you can add --all argument to get extended information, like transcript sequences.
kmerator --info MLLT3 ENSG00000157764 # info about the MLLT3 and BRAF genes
kmerator --info mllt3 -r 109 -S mouse # info about mllt3 against mus_musculus release 109
kmerator --info MMLT3 --all # extended info about the MMLT3 gene
kmerator --info genes.txt # genes/transcripts are in a file
options:
-h, --help show this help message and exit
-s SELECTION [SELECTION ...], --selection SELECTION [SELECTION ...]
list of gene IDs (ENSG, gene Symbol or alias) or transcript
IDs (ENST) from which you want to extract specific kmers from.
For genes, kmerator search specific kmers along the gene.
For transcripts, it search specific kmers to the transcript.
You can also give a file with yours genes/transcripts
separated by space, tab or newline. If you want to use
your own unannotated sequences, you must give your fasta
file with --fasta-file option.
-f FASTA_FILE, --fasta-file FASTA_FILE
Use this option when yours sequences are unannonated or
provided by a annotation file external from Ensembl.
Otherwise, use --selection option.
-d DATADIR, --datadir DATADIR
Storage directory for kmerator datasets.We recommend to
set this parameter by editing the configuration file
(kmerator --edit)
-g GENOME, --genome GENOME
Genome jellyfish index (.jf) to use for k-mers requests.
-S SPECIE, --specie SPECIE
indicate a specie referenced in Ensembl, to help, follow
the link https://rest.ensembl.org/documentation/info/species.
You can use the 'name', the 'display_name' or any 'alias'.
For example human or homo_sapiens are valid
(default:human).
-k KMER_LENGTH, --kmer-length KMER_LENGTH
k-mer length that you want to use (default 31).
-r RELEASE, --release RELEASE
release of transcriptome (default: last).
--stringent Only for genes with '-s/--selection' option: use this
option if you want to select gene-specific k-mers present
in ALL transcripts for your gene. If false, a k-mer is
considered as gene-specific if present in at least one
isoform of your gene of interest.
-T MAX_ON_TRANSCRIPTOME, --max-on-transcriptome MAX_ON_TRANSCRIPTOME
Only for genes with '-f/--fasta-file' option: with
unanotated data, specific kmers are not supposed to be
found in the transcriptome, you can change this behavior
for special cases (default: 0)
-G MAX_ON_GENOME, --max-on-genome MAX_ON_GENOME
Only for genes with '-f/--fasta-file' option: typically,
specific kmers are not supposed to be found more than
once in the genome, you can change
this behavior for special cases, like chimera (default: 1)
-o OUTPUT, --output OUTPUT
output directory, created if not exists (default: 'output')
-t THREAD, --thread THREAD
run n process simultaneously (default: 1)
--tmpdir TMPDIR directory to temporary file (default: /tmp/kmerator_<random>)
-D, --debug Show more details while Kmerator is running.
--keep keep intermediate files (sequences, indexes, separate
kmers and contigs files).
-y, --yes assumes 'yes' as the prompt answer, run non-interactively.
-e, --edit-config Edit config file
-l, --list-dataset, --list-datasets
list the local datasets (based on the datadir option).
--rm-dataset remove a dataset, according with --specie and
--release options
--mk-dataset make a dataset, according with --specie and --release options
--last-avail, --last-available
last release available on Ensembl
-u, --update-dataset builds a new dataset if a new version is found on Ensembl
--info gene [gene ...]
get some information about gene. Multiple entries are
allowed or a text file with list of genes
-a, --all only with '--info' option. Give more info, like
transcript sequences
-v, --version show program's version number and exit
kmerator genera 6246 te 3 files:
- kmers.fa
- contigs.fa
- report.md
extract of kmers.fa:
>RUNX1:ENST00000675419.kmer3394 ct:3 tr:3/17
TGAAGAGTATTTGAAAGCAGGACTTCAGAAC
kmer3394: the first base is at this position in the canonical transcript or the given sequence (1 based)ct:3: the kmer is associated to the 3rd contig of this gene/transcript, relative tocontigs.fafileex3/17: the kmer is located in the 3rd exon of the 17
extract of contigs.fa:
>RUNX1:ENST00000675419.contig_3 (at position 2314)
ACTTCTTTGGGCCTCATAAACAACCACAGAACCACAAGTTGGGTAGCCTGGCAGTGTCAGAAGTCTGAACCCAG
contig_3: contig count, relative toct:3ofkmers.faheadersposition 2314: the first base of the contig is at this position (1 based)
[1] Guillaume Marçais, Carl Kingsford, A fast, lock-free approach for efficient parallel counting of occurrences of k-mers, Bioinformatics, Volume 27, Issue 6, 15 March 2011, Pages 764–770, https://doi.org/10.1093/bioinformatics/btr011 [2] Rodriguez JM, et al. Nucleic Acids Res. Database issue; 2017 Oct 23