Skip to content

Plasmid Atlas - A web interface to browse for plasmids and their associated genes. Visit us at:

License

Notifications You must be signed in to change notification settings

tiagofilipe12/pATLAS

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

pATLAS logo

Join the chat at https://gitter.im/plasmidATLAS/Lobby Codacy Badge

Table of contents

Description

Plasmid Atlas is a web-base tool that empowers researchers to easily and rapidly access information related with plasmids present in NCBI's refseq database. In pATLAS each node (or circle) represents a plasmid and each link between two plasmids means that those two plasmids share around 90% average nucleotide identity.

With this tool we have two main goals:

  1. Increase the accessibility of plasmid relevant metadata to users as well as facilitate the access to that metadata.
  2. Improve the ease of interpreting results from High Throughput Sequencing (HTS) for plasmid detection.

Citation

Tiago F Jesus, Bruno Ribeiro-Gonçalves, Diogo N Silva, Valeria Bortolaia, Mário Ramirez, João A Carriço; Plasmid ATLAS: plasmid visual analytics and identification in high-throughput sequencing data, Nucleic Acids Research, gky1073, https://doi.org/10.1093/nar/gky1073

Documentation

If are interested in learning how to use pATLAS, please refer to gitbook documentation.


Development

Dependencies

  • Mash (2.0) - You can download mash version 2.0.0 directly here: linux and OSX. Other releases were not tested but may be downloaded in Mash git releases page.

  • Postgresql (>= 10.0) - This script uses Postgres database to store the database: releases page

  • Python 3 and the respective pip.

  • To install all other dependencies just run: pip install -r requirements.txt

Backend Scripts

MASHix.py

MASHix.py is the main script to generate the database. This script generates a matrix of pairwise comparisons between sequences in input fasta(s) file(s). Note that it reads multifastas, i.e., each header in fasta is a reference sequence.

Options:

Main options:
'-i','--input_references' - 'Provide the input fasta files to parse.
                            This will inputs will be joined in a
                            master fasta.'

'-o','--output' - 'Provide an output tag'

'-t', '--threads' - 'Provide the number of threads to be used'

'-db', '--database_name' - 'This argument must be provided as the last
argument. It states the database name that must be used.'
MASH related options:
'-k','--kmers' - 'Provide the number of k-mers to be provided to mash
                sketch. Default: 21'

'-p','--pvalue' - 'Provide the p-value to consider a distance
                significant. Default: 0.05'

'-md','--mashdist' - 'Provide the maximum mash distance to be
                    parsed to the matrix. Default:0.1'
Other options:
'-no_rm', '--no-remove' - 'Specify if you do not want to remove the
                        output concatenated fasta.'

'-hist', '--histograms' - 'Checks the distribution of distances
                        values ploting histograms.'

'-non', '--nodes_ncbi' - 'specify the path to the file containing
                        nodes.dmp from NCBI'

'-nan', '--names_ncbi' - 'specify the path to the file containing
                        names.dmp from NCBI'

'--search-sequences-to-remove' - 'this option allows to only run the
                                 part of the script that is required
                                 to generate the filtered fasta.
                                 Allowing for instance to debug
                                 sequences that shoudn't be removed
                                 using 'cds' and 'origin' keywords'.

Database customization:

I don't like database name! How do I change it?

Go to db_manager/config_default.py and edit the following line:

SQLALCHEMY_DATABASE_URI = 'postgresql:///<custom_database_name>'
I don't like table name inside database! How do I change it?

Go to db_manager/db_app/models.py and edit the following line:

 __tablename__ = "<custom_table_name>"

Database migration from one server to another

Database export
pg_dump <db_name> > <file_name.sql>
Database import
psql -U <user_name> -d <db_name> -f <file_name.sql>

Supplementary scripts

abricate2db.py

This script inherits a class from ODiogoSilva/Templates and uses it to parse abricate outputs and dumps abricate outputs to a psql database, depending on the input type provided.

Options:

"-i", "--input_file" - "Provide the abricate file to parse to db.
                        It can accept more than one file in the case of
                        resistances."

"-db_psql", "--database_name" - "his argument must be provided as the
                                last argument. It states the database
                                name that must be used."

"-db", "--db" - "Provide the db to output in psql models."

"-id", "--identity" - "minimum identity to be reported to db"

"-cov", "--coverage" - "minimum coverage do be reported to db"

"-csv", "--csv" - "Provide card csv file to get correspondence between
                    DNA accessions and ARO accessions. Usually named
                    aro_index.csv. By default this file is already
                    available in patlas repo with a specific path:
                    'db_manager/db_app/static/csv/aro_index.csv'"

taxa_fetch.py

This script is located in utils folder and can be used to generate a JSON file with the corresponding taxonomic tree. It fetches for a given species, the genera, family and order to which it belongs. Note: for plasmids I have to make some filtering in the resulting taxids and list of species that other users may want to skip

Options:

-i INPUT_LIST, --input_list INPUT_LIST
                        provide a file with a listof species. Each
                        speciesshould be in each line.
-non NODES_FILE, --nodes_ncbi NODES_FILE
                        specify the path to the file containing
                        nodes.dmp from NCBI
-nan NAMES_FILE, --names_ncbi NAMES_FILE
                        specify the path to the file containing
                        names.dmp from NCBI
-w, --weirdos         This option allows the userto add a checks for
                        weirdentries. This is mainly usedto parse the
                        plasmids refseq, so if you do not want this to
                        be used, use this option.

List of entries that will be filtered from weirdos option
  • From taxonomy levels:

    • "bug"
    • "insect"
    • "angiosperm"
    • "fungus"
    • "cnidarian"
    • "mudpuppy"
    • "mantid"
    • "mussel"
  • From species in fasta headers:

    • "orf"
    • "unknown"
    • "Uncultured"
    • "uncultured"
    • "Peanut"
    • "Pigeon"
    • "Wheat"
    • "Beet"
    • "Blood"
    • "Onion"
    • "Tomato"
    • "Zea"
    • "Endosymbiont"
    • "Bacillaceae"
    • "Comamonadaceae"
    • "Enterobacteriaceae"
    • "Opitutaceae"
    • "Rhodobacteraceae"
    • "Bacterium"
    • "Endophytic"
  • It also attempts to fix some bugs in species naming like the following:

    • "B bronchiseptica"
    • "S pyogenes"

Note: Yes people like to give interesting names to bacteria...

pATLAS API

Schematics of the pATLAS database creation

Workflow db creation

Workflow for database creation

  1. Download plasmid sequences available in NCBI refseq
  2. Extract fasta from tar.gz
  3. Download and extract NCBI taxonomy, which will be fed to pATLAS.
  4. Clone this repository:
git clone https://github.com/tiagofilipe12/pATLAS
  1. Install its dependencies

  2. Configure the database:

createdb <database_name>
pATLAS/patlas/db_manager/db_create.py <database_name>
  1. run MASHix.py - the output will include a filtered fasta file (master_fasta_*.fas).
  2. run ABRicate, with CARD, ResFinder, PlasmidFinder, VFDB databases.
# e.g.
abricate --db card <master_fasta*.fas> > abr_card.tsv
abricate --db resfinder <master_fasta*.fas> > abr_resfinder.tsv
abricate --db vfdb <master_fasta*.fas> > abr_vfdb.tsv
abricate --db plasmidfinder <master_fasta*.fas> > abr_plasmidfinder.tsv
  1. Download the card index necessary for the abricate2db.py script (aro_index.csv).
  2. run abricate2db.py - using all the previous tsv as input.
# e.g.
abricate2db.py -i abr_plasmidfinder.tsv -db plasmidfinder \
    -id 80 -cov 90 -csv aro_index.csv -db_psql <database_name>
  1. dump database to a sql file.

Automation of this steps

This steps are fully automated in the nextflow pipeline pATLAS-db-creation.

Creating a custom version of pATLAS

If you require to add your own plasmids to pATLAS database without asking to add them to pATLAS website, you can provide custom fasta files when building the database using the -i option of MASHix.py. Then follow the steps described above.

Run pATLAS locally

Docker compose

You can run pATLAS locally without much requirements by using patlas-compose. This will automatically handle the installation of the version 1.5.2 of pATLAS and launch the service in a local instance. For that you just require:

Then, follow this simple steps:

git clone https://github.com/bfrgoncalves/patlas-compose
  • Enter the patlas-compose folder
cd patlas-compose
  • Launch the compose:
docker-compose up
  • Wait for the line * Running on http://0.0.0.0:5000/ (Press CTRL+C to quit) to show up, meaning that the service is now running.

  • Access on 127.0.0.1:5000 or 0.0.0.0:5000.

Note: This methodology is highly recommended.

From scratch

pATLAS can be run locally if you have PostgreSQL installed and configured. After, you just need to:

  1. Clone this repository:
git clone https://github.com/tiagofilipe12/pATLAS
  1. Create your custom database version or generate the default pATLAS database or download sql file from version 1.5.2 (the tar.gz archive). Note: if you download the sql file from version 1.5.2 you may skip steps 3 to 4 and continue with step 5.

  2. Make sure all the necessary files are in place.

  • by default pATLAS generates a import_to_vivagraph.json file in the folder <tag_provided_to_o_flag>/results. Place this file in the patlas/db_manager/db_app/static/json folder.
  • change session to read the new import_to_vivagraph.json file by changing from false to true a variable named devel in patlas/db_manager/db_app/static/js/pATLASGlobals.js
  1. Create the database that the front end will run:
createdb <your_database>
  1. load the generated sql file

  2. Install backend dependencies:

# within the root directory of this repository
pip install -r requirements.txt
  1. Install frontend dependencies:
# change directory to static direcoty where `index.html` will look for
# its depdendenies
cd patlas/db_manager/db_app/static/
# then install them (package.json is located in this directory)
yarn install
  1. Compile node modules so that the html can understand, using webpack:
# You can also user a local installation of webpack.
# entry-point.js is the config file where all the imported modules are
# called
node_modules/webpack/bin/webpack.js entry-point.js
  1. Then execute the script run.py.
# within the root directory of this repository
cd patlas/db_manager
./run.py <your_database>

Note: the database name is utterly important to properly say to the frontend where to get the data.

  1. Go to 127.0.0.1:5000.

Optimization of the resources usage by the web page

Using the devel = true isn't very efficient, so you can allow the force directed graph to render in a devel = true session, then when you are satisfied pause the force layout using the buttons available in pATLAS and click at the same time Shift+Ctrl+Space. This will take a while but eventually it will generate a file named filtered.json. Once you have this file you can add it to the patlas/db_manager/db_app/static/json folder and change the devel variable to false. This will use the previously saved positions to render a pre rendered network.