Multiprocessing for Pathway Tools
mpwt: Multiprocessing Pathway Tools
mpwt is a python package for running Pathway Tools [Karp2019] on multiple genomes using multiprocessing. More precisely, it launches one PathoLogic [Karp2011] process for each organism. This allows to increase the speed of draft metabolic network reconstruction when working on multiple organisms.
There is no guarantee that this script will work, it is a Work In Progress in early state.
mpwt: Pipeline summary
The following picture shows the main argument of mpwt:
Table of contents
- For developer
- Release Notes
You must have an environment where Pathway Tools is installed. Pathway Tools can be obtained here. The last version supported by mpwt is shown in the badge Pathway Tools.
Pathway Tools needs Blast, so it must be install on your system. Depending on your system, Pathway Tools needs a file named .ncbirc to locate Blast, for more informations look at this page.
/!\ For all OS, Pathway-Tools must be in $PATH.
On Linux and MacOS: export PATH=$PATH:your/install/directory/pathway-tools.
Consider adding Pathway Tools in $PATH permanently by running:
echo 'export PATH="$PATH:your/install/directory/pathway-tools:"' >> ~/.bashrc
If your OS doesn’t support Pathway Tools, you can use a docker container. If it’s your case, look at Pathway Tools Multiprocessing Docker. It is a dockerfile that will create a container with Pathway Tools, its dependencies and this package. You just need to give a Pathway Tools installer as input.
You can also look at Pathway Tools Multiprocessing Singularity. More manipulations are required compared to Docker but with this you can create a Singularity image.
pip install mpwt
The script takes a folder containing sub-folders as input. Each sub-folder contains a Genbank/GFF file or multiple PathoLogic Format (PF) files.
Folder_input ├── species_1 │ └── species_1.gbk ├── species_2 │ └── species_2.gff │ └── species_2.fasta ├── species_3 │ └── species_3.gbk ├── species_4 │ └── scaffold_1.pf │ └── scaffold_1.fasta │ └── scaffold_2.pf │ └── scaffold_2.fsa taxon_id.tsv ..
Input files must have the same name as the folder in which they are located and also finished with a .gbk/.gbff or a .gff.
For PF files, there is one file for each scaffold/contig and one corresponding fasta file.
Pathway Tools will run on each Genbank/GFF/PF files. It will create the results in the ptools-local folder but you can also choose an output folder.
Folder_input ├── species_1 │ └── species_1.gbk ..
Genbank file example:
LOCUS scaffold1 XXXXXX bp DNA linear INV DD-MMM-YYYY DEFINITION My species genbank. ACCESSION scaffold1 VERSION scaffold1 KEYWORDS Key words. SOURCE Source ORGANISM Species name Taxonomy; Of; My; Species; With; The; Genus. FEATURES Location/Qualifiers source 1..XXXXXX /scaffold="scaffold1" /db_xref="taxon:taxonid" gene START..STOP /locus_tag="gene1" mRNA START..STOP /locus_tag="gene1" CDS START..STOP /locus_tag="gene1" /db_xref="InterPro:IPRXXXXXX" /go_component="GO:XXXXXXX" /EC_number="X.X.X.X" /translation="AMINOAACIDSSEQUENCE"
Folder_input ├── species_2 │ └── species_2.gff │ └── species_2.fasta ..
GFF file example:
##gff-version 3 ##sequence-region scaffold_1 1 XXXXXX scaffold_1 RefSeq region 1 XXXXXXX . + . ID=region_id;Dbxref=taxon:XXXXXX scaffold_1 RefSeq gene START STOP . - . ID=gene_id scaffold_1 RefSeq CDS START STOP . - 0 ID=cds_id;Parent=gene_id;ec_number=X.X.X.X"
Warning: it seems that metabolic networks from GFF file have less reactions/pathways/compounds than metabolic networks from Genbank file or PathoLogic File. Lack of some annotations (EC, GO) can be the reason explaining these differences.
Look at the NCBI GFF format for more informations.
You have to provide a nucleotide sequence file (either ‘.fasta’ or ‘.fsa’ extensions) associated with the GFF file containing the chromosome/scaffold/contig sequence.
Folder_input ├── species_4 │ └── scaffold_1.pf │ └── scaffold_1.fasta │ └── scaffold_2.pf │ └── scaffold_2.fsa taxon_id.tsv ..
PF file example:
;;;;;;;;;;;;;;;;;;;;;;;;; ;; scaffold_1 ;;;;;;;;;;;;;;;;;;;;;;;;; ID gene_id NAME gene_id STARTBASE START ENDBASE STOP FUNCTION ORF PRODUCT-TYPE P PRODUCT-ID prot gene_id EC X.X.X.X DBLINK GO:XXXXXXX INTRON START1-STOP1 //
Look at the Pathologic format for more informations.
You have to provide one nucleotide sequence (either ‘.fasta’ or ‘.fsa’ extension) for each pathologic containing one scaffold/contig.
Also to add the taxon ID we need the taxon_id.tsv (a tsv file with two values: the name of the folder containing the PF files and the taxon ID corresponding).
If you don’t have taxon ID in your Genbank or GFF file, you can add one in this file for the corresponding species.
You can also add more informations for the genetic elements like circularity of genome (Y or N), type of genetic element (:CHRSM, :PLASMID, :MT (mitochondrial chromosome), :PT (chloroplast chromosome), or :CONTIG) or codon table (see the corresponding code below).
As you can see for PF file (species_4) you can use the column corresponding_file to add information for each PF files.
Genetic code for Pathway Tools:
|Corresponding number||Genetic code|
|1||The Standard Code|
|2||The Vertebrate Mitochondrial Code|
|3||The Yeast Mitochondrial Code|
|4||The Mold, Protozoan, and Coelenterate Mitochondrial Code and the Mycoplasma/Spiroplasma Code|
|5||The Invertebrate Mitochondrial Code|
|6||The Ciliate, Dasycladacean and Hexamita Nuclear Code|
|9||The Echinoderm and Flatworm Mitochondrial Code|
|10||The Euplotid Nuclear Code|
|11||The Bacterial, Archaeal and Plant Plastid Code|
|12||The Alternative Yeast Nuclear Code|
|13||The Ascidian Mitochondrial Code|
|14||The Alternative Flatworm Mitochondrial Code|
|15||Blepharisma Nuclear Code|
|16||Chlorophycean Mitochondrial Code|
|21||Trematode Mitochondrial Code|
|22||Scenedesmus obliquus Mitochondrial Code|
|23||Thraustochytrium Mitochondrial Code|
Three input files are created by mpwt. Informations are extracted from the Genbank/GFF/PF file. myDBName corresponds to the name of the folder and the Genbank/GFF/PF file. taxonid corresponds to the taxonid in the db_xref of the source feature in the Genbank/GFF/PF. The species_name is extracted from the Genbank/GFF/PF files.
**organism-params.dat** ID myDBName STORAGE FILE NCBI-TAXON-ID taxonid NAME species_name **genetic-elements.dats** NAME ANNOT-FILE gbk_pathname // **dat_creation.lisp** (in-package :ecocyc) (select-organism :org-id 'myDBName) (let ((*progress-noter-enabled?* NIL)) (create-flat-files-for-current-kb))
By using the python multiprocessing library, mpwt launches parallel PathoLogic processes on physical cores. Regarding memory requirements, they depend on the genome but we advise to use at least 2 GB per core.
mpwt can be used with the command line:
mpwt -f path/to/folder/input [-o path/to/folder/output] [--patho] [--hf] [--op] [--tp] [--nc] [-p FLOAT] [--dat] [--md] [--cpu INT] [-r] [--clean] [--log path/to/folder/log] [--ignore-error] [-v]
Optional argument are identified by .
mpwt can be used in a python script with an import:
import mpwt folder_input = "path/to/folder/input" folder_output = "path/to/folder/output" mpwt.multiprocess_pwt(input_folder=folder_input, output_folder=folder_output, patho_inference=optional_boolean, patho_hole_filler=optional_boolean, patho_operon_predictor=optional_boolean, patho_transporter_inference=patho_transporter_inference, no_download_articles=optional_boolean, dat_creation=optional_boolean, dat_extraction=optional_boolean, size_reduction=optional_boolean, number_cpu=int, patho_log=optional_folder_pathname, ignore_error=optional_boolean, pathway_score=pathway_score, taxon_file=optional_boolean, verbose=optional_boolean)
|Command line argument||Python argument||description|
|-f||input_folder(string: folder pathname)||Input folder as described in Input data|
|-o||output_folder(string: folder pathname)||Output folder containing PGDB data or dat files (see –dat arguments)|
|–patho||patho_inference(boolean)||Launch PathoLogic inference on input folder|
|–hf||patho_hole_filler(boolean)||Launch PathoLogic Hole Filler with Blast|
|–op||patho_operon_predictor(boolean)||Launch PathoLogic Operon Predictor|
|–tp||patho_transporter_inference(boolean)||Launch PathoLogic Transport Inference Parser|
|–nc||no_download_articles(boolean)||Launch PathoLogic without loading PubMed citations (not working)|
|-p||pathway_score(float)||Launch PathoLogic using a specified pathway prediction score cutoff|
|–dat||dat_creation(boolean)||Create BioPAX/attribute-value dat files|
|–md||dat_extraction(boolean)||Move only the dat files inside the output folder|
|–cpu||number_cpu(int)||Number of cpu used for the multiprocessing|
|-r||size_reduction(boolean)||Delete PGDB in ptools-local to reduce size and return compressed files|
|–log||patho_log(string: folder pathname)||Folder where log files for PathoLogic inference will be store|
|–delete||mpwt.remove_pgdbs(string: pgdb name)||Delete a specific PGDB|
|–clean||mpwt.cleaning()||Delete all PGDBs in ptools-local folder or only PGDB from input folder|
|–ignore-error||ignore_error(boolean)||Ignore errors and continue the workflow for successful build|
|–taxon-file||taxon_file(boolean)||Force mpwt to use the taxon ID in the taxon_id.tsv file|
|-v||verbose(boolean)||Print some information about the processing of mpwt|
There is also another argument:
mpwt topf -f input_folder -o output_folder --cpu cpu_number
import mpwt mpwt.to_pathologic.create_pathologic_file(input_folder, output_folder, cpu_number)
This argument reads the input data inside the input folder. Then it converts Genbank and GFF files into PathoLogic Format files. And if there is already PathoLogic files it copies them.
It can be used to avoid issues with parsing Genbank and GFF files. But it is an early Work in Progress.
The –hf/patho_hole_filler option uses the Hole Filler [Karp2019arXiv]:
The pathway hole-filling program PHFiller (a component of PathoLogic) generates hypotheses as to which genes code for these missing enzymes by using the following method. Given a reaction that is a pathway hole, the program first queries the UniProt database to find all known sequences for enzymes that catalyze that same reaction in other organisms. The program then uses the BLAST tool to compare that set of sequences against the full proteome of the organism in which we are seeking hole fillers. It scores the resulting BLAST hits using a Bayesian classifier that considers information such as genome localization (that is, is a potential hole filler in the same operon as another gene in the same metabolic pathway?). At a stringent probability-score cutoff, our method finds potential hole fillers for approximately 45% of the pathway holes in a microbial genome .
This option is more precisely described in [Green2004]:
- Sequence retrieval – Retrieve from Swiss-Prot and PIR sequences for enzymes that catalyze the desired reaction in other organisms. Because these sequences are not necessarily homologs, we will refer to enzymes with the same function in a variety of organisms as isozymes. For Swiss-Prot, the program retrieves Swiss-Prot IDs directly from the ENZYME database. For PIR sequences, the program retrieves IDs from the MetaCyc PGDB. Sequences are then retrieved directly from the most recent version of each database.
- Homology search – BLAST each query isozyme sequence against the genome of the organism of interest.
- Data consolidation – Congruence analysis of the resulting BLAST hits to consolidate data reported for sequences that align with one or more query isozymes.
- Candidate evaluation – Determine the probability that each candidate protein has the activity required by the missing reaction.
The –op/patho_operon_predictor identifies operon [Karp2019arXiv]:
The Pathway Tools operon predictor identifies operon boundaries by examining pairs of adjacent genes A and B and using information such as intergenic distance, and whether it can identify a functional relationship between A and B, such as membership in the same pathway, membership in the same multimeric protein complex, or whether A is a transporter for a substrate within a metabolic pathway in which B is an enzyme.
The –tp/patho_transporter_inference tries to answer the question “What chemicals can the organism import or export?” [Karp2019arXiv]:
To answer such queries, Pathway Tools uses an ontology-based representation of transporter function in which transport events are represented as reactions in which the transported compound(s) are substrates. Each substrate is labeled with the cellular compartment in which it resides, and each substrate is a controlled-vocabulary term from the extensive set of chemical compounds in MetaCyc. The TIP program converts the free-text descriptions of transporter functions found in genome annotations (examples: “phosphate ABC transporter”and “sodium/proline symporter”) into computable transport reactions.
The -p/pathway_score determines the cutoff for pathway prediction.
This cutoff is defined in ptools-init.dat:
During the pathway prediction process, pathways are assigned a score between 0 and 1 based on the evidence for the presence of that pathway. Pathways whose score does not exceed this cutoff value will usually be rejected (although certain rules may cause them to be predicted as present).
This pathway prediction score has also been explained in [Karp2018]:
A very strict pathway score cutoff of 1.0 was supplied to PathoLogic to predict into BlongCyc (from MetaCyc) only the pathways that have gene annotations associated with all pathway reactions, to minimize the effects of pathway inference on biomass goal reachability. PathoLogic inference of a metabolic pathway causes all reactions within the pathway to be imported from the MetaCyc database into the new PGDB, including reactions lacking gene assignments — using the 1.0 cutoff means that no reactions lacking gene assignments were imported from MetaCyc during pathway inference. The resulting PGDB was subjected to the following manual refinement steps. That is, some manual refinement occurred before gap filling began.
Possible uses of mpwt:
command lineimport mpwt python script
Create PGDBs of studied organisms inside ptools-local:
mpwt -f path/to/folder/input --pathoimport mpwt mpwt.multiprocess_pwt(input_folder='path/to/folder/input', patho_inference=True)
Convert Genbank and GFF files into PathoLogic files then create PGDBs of studied organisms inside ptools-local:
mpwt topf -f path/to/folder/input -o path/to/folder/pf mpwt -f path/to/folder/pf --pathoimport mpwt mpwt.create_pathologic_file(input_folder='path/to/folder/input', output_folder='path/to/folder/pf') mpwt.multiprocess_pwt(input_folder='path/to/folder/pf', patho_inference=True)
Create PGDBs of studied organisms inside ptools-local with Hole Filler, Operon Predictor, Transport Inference Parser and create logs:
mpwt -f path/to/folder/input --patho --hf --op --tp --log path/to/folder/logimport mpwt mpwt.multiprocess_pwt(input_folder='path/to/folder/input', patho_inference=True, patho_hole_filler=True, patho_operon_predictor=True, patho_transporter_inference=True, patho_log='path/to/folder/log')
Create PGDBs of studied organisms inside ptools-local with pathway prediction score of 1:
mpwt -f path/to/folder/input --patho -p 1.0import mpwt mpwt.multiprocess_pwt(input_folder='path/to/folder/input', patho_inference=True, pathway_score=1.0)
Create PGDBs of studied organisms inside ptools-local and create dat files:
mpwt -f path/to/folder/input --patho --datimport mpwt mpwt.multiprocess_pwt(input_folder='path/to/folder/input', patho_inference=True, dat_creation=True)
Create PGDBs of studied organisms inside ptools-local. Then move all the PGDB files to the output folder.
mpwt -f path/to/folder/input --patho -o path/to/folder/outputimport mpwt mpwt.multiprocess_pwt(input_folder='path/to/folder/input', output_folder='path/to/folder/output', patho_inference=True)
Create PGDBs of studied organisms inside ptools-local and create dat files. Then move the dat files to the output folder.
mpwt -f path/to/folder/input --patho --dat -o path/to/folder/output --mdimport mpwt mpwt.multiprocess_pwt(input_folder='path/to/folder/input', output_folder='path/to/folder/output', patho_inference=True, dat_creation=True, dat_extraction=True)
Create dat files for the PGDB inside ptools-local. And move them to the output folder.
mpwt --dat -o path/to/folder/output --mdimport mpwt mpwt.multiprocess_pwt(output_folder='path/to/folder/output', dat_creation=True, dat_extraction=True)
Move PGDB from ptools-local to the output folder:
mpwt -o path/to/folder/outputimport mpwt mpwt.multiprocess_pwt(output_folder='path/to/folder/output')
Move dat files from ptools-local to the output folder:
mpwt -o path/to/folder/output --mdimport mpwt mpwt.multiprocess_pwt(output_folder='path/to/folder/output', dat_extraction=True)
- Run the multiprocess Pathway Tools on input folder
import mpwt mpwt.multiprocess_pwt(input_folder=folder_input, output_folder=folder_output, patho_inference=optional_boolean, patho_hole_filler=optional_boolean, patho_operon_predictor=optional_boolean, patho_transporter_inference=patho_transporter_inference, no_download_articles=optional_boolean, dat_creation=optional_boolean, dat_extraction=optional_boolean, size_reduction=optional_boolean, number_cpu=int, patho_log=optional_folder_pathname, ignore_error=optional_boolean, pathway_score=pathway_score, taxon_file=optional_boolean, verbose=optional_boolean)
- Delete all the previous PGDB and the metadata files
import mpwt mpwt.cleaning(number_cpu=optional_int, verbose=optional_boolean)
This can also be used with a command line argument:mpwt --clean
If you use clean and the argument -f input_folder, it will delete input files (‘dat_creation.lisp’, ‘dat_creation.log’, ‘pathologic.log’, ‘pwt_terminal.log’, ‘genetic-elements.dat’ and ‘organism-params.dat’) and the PGDB corresponding to the input folder.mpwt -f input_folder --clean
For example if you have:Folder_input ├── species_1 │ └── species_1.gbk ├── species_2 │ └── species_2.gff │ └── species_2.fasta ├── species_3 │ └── species_3.gbk
And you have in your ptools-local:ptools-local ├── pgdbs ├── user ├── species_1cyc │ └── .. ├── species_2cyc │ └── .. ├── species_3cyc │ └── .. ├── species_4cyc │ └── ..
The command:mpwt -f input_folder --clean
will delete species_1cyc, species_2cyc and species_3cyc but not species_4cyc.
- Delete a specific PGDB
With this command, it is possible to delete a specific PGDB, where pgdb_name is the name of the PGDB (ending with ‘cyc’). It can be multiple pgdbs, to do this, put all the pgdb IDs in a string separated by a ‘,’.import mpwt mpwt.remove_pgdbs(pgdb_name)
And as a command line:mpwt --delete mydbcyc1,mydbcyc2
- Return the path of ptools-local
import mpwt ptools_local_path = mpwt.find_ptools_path()
- Return a list containing all the PGDBs inside ptools-local folder
import mpwt list_of_pgdbs = mpwt.list_pgdb()
Can be used as a command with:mpwt --list
If you encounter errors (and it is highly possible) there is informations that can help you resolved them.
For error during PathoLogic inference, you can use the log arguments. The log contains the summary of the build and the error for each species. There is also a pathologic.log (created by Pathway Tools), a pwt_terminal.log (log of the terminal during PathoLogic process) and a dat_creation.log (log of the terminal during attributes-values files creation) in each sub-folders.
If the build passed you have also the possibility to see the result of the inference with the file resume_inference.tsv. For each species, it contains the number of genes/proteins/reactions/pathways/compounds in the metabolic network.
If Pathway Tools crashed, mpwt can print some useful information in verbose mode. It will show the terminal in which Pathway Tools has crashed. Also, if there is an error in pathologic.log, it will be shown after === Error in Pathologic.log ===.
There is a Pathway Tools forum where you can find informations on Pathway Tools errors.
You can also ignore PathoLogic errors by using the argument –ignore-error/ignore_error. This option will ignore error and continue the mpwt workflow on the successful PathoLogic build.
If you did not use the output argument, results (PGDB with/without BioPAX/dat files) will be inside your ptools-local folder ready to be used with Pathway Tools. Have in mind that mpwt does not create the cellular overview and does not used the hole-filler. So if you want these results you should run them after.
If you used the output argument, there is two potential outputs depending on the use of the option –md/dat_extraction:
- without –md/dat_extraction, you will have a complete PGDB folder inside your results, for example:
Folder_output ├── species_1 │ └── default-version │ └── 1.0 │ └── data │ └── contains BioPAX/dat files if you used the --dat/dat_creation option. │ └── input │ └── species_1.gbk │ └── genetic-elements.dat │ └── organism-init.dat │ └── organism.dat │ └── kb │ └── species_1.ocelot │ └── reports │ └── contains Pathway Tools reports. ├── species_2 .. ├── species_3 ..
- with –md/dat_extraction, you will only have the dat files, for example:
Folder_output ├── species_1 │ └── classes.dat │ └── compounds.dat │ └── dnabindsites.dat │ └── enzrxns.dat │ └── genes.dat │ └── pathways.dat │ └── promoters.dat │ └── protein-features.dat │ └── proteins.dat │ └── protligandcplxes.dat │ └── pubs.dat │ └── reactions.dat │ └── regulation.dat │ └── regulons.dat │ └── rnas.dat │ └── species.dat │ └── terminators.dat │ └── transunits.dat │ └── .. ├── species_2 .. ├── species_3 ..
- with the -r /size_reduction argument, you will have compressed zip files (and PGDBs inside ptools-local will be deleted):
Folder_output ├── species_1.zip ├── species_2.zip ├── species_3.zip ..
mpwt uses logging so you need to create the handler configuration if you want mpwt’s log in your application:
import logging from mpwt import multiprocess_pwt logging.basicConfig() multiprocess_pwt(...)
|[Green2004]||Green, M.L., Karp, P.D. A Bayesian method for identifying missing enzymes in predicted metabolic pathway databases. BMC Bioinformatics 5, 76 (2004). https://doi.org/10.1186/1471-2105-5-76|
|[Karp2011]||Karp, P. D., Latendresse, M., & Caspi, R. The pathway tools pathway prediction algorithm. Standards in genomic sciences 5(3), 424–429 (2011). https://doi.org/10.4056/sigs.1794338|
|[Karp2018]||Karp, P. D., Weaver, D. & Latendresse, M. How accurate is automated gap filling of metabolic models?. BMC Systems Biology 12(1), 73 (2018). https://doi.org/10.1186/s12918-018-0593-7|
|[Karp2019arXiv]||(1, 2, 3) Karp, P. D., Paley, S. M., Midford, P. E., Krummenacker, M., Billington, R., Kothari, A., Ong, W. K., Subhraveti, P., Keseler, I. M. & Caspi R. Pathway Tools version 23.0: Integrated Software for Pathway/Genome Informatics and Systems Biology. arXiv (2019). https://arxiv.org/abs/1510.03964|
|[Karp2019]||Karp, P. D., Midford, P. E., Billington, R., Kothari, A., Krummenacker, M., Latendresse, M., Ong, W. K., Subhraveti, P., Caspi, R., Fulcher, C., Keseler, I. M., & Paley, S. M. Pathway Tools version 23.0 update: software for pathway/genome informatics and systems biology. Briefings in bioinformatics, bbz104. Advance online publication (2019). https://doi.org/10.1093/bib/bbz104|
Arnaud Belcour, Clémence Frioux, Meziane Aite, Anthony Bretaudeau, Anne Siegel (2019) Metage2Metabo: metabolic complementarity applied to genomes of large-scale microbiotas for the identification of keystone species. bioRxiv 803056; doi: https://doi.org/10.1101/803056.
Mézaine Aite for his work on the first draft of this package.
Clémence Frioux for her work and feedbacks.
Peter Karp, Suzanne Paley, Markus Krummenacker, Richard Billington and Anamika Kothari from the Bioinformatics Research Group of SRI International for their help on Pathway Tools and on Genbank format.
GenOuest bioinformatics (https://www.genouest.org/) core facility for providing the computing infrastructure to test this tool.
All the users that have tested this tool.
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size mpwt-0.5.9-py3-none-any.whl (48.8 kB)||File type Wheel||Python version py3||Upload date||Hashes View|
|Filename, size mpwt-0.5.9.tar.gz (52.5 kB)||File type Source||Python version None||Upload date||Hashes View|