tl;dr - download and sketch data directly
Commands:
gbsketch- download and sketch NCBI Assembly Datasets by accessionurlsketch- download and sketch directly from a url
This plugin is an attempt to improve sourmash database generation by downloading files, checking md5sum if provided, and sketching to a sourmash zipfile. FASTA/Q files can also be saved if desired. It's under active development.
Option 1 (recommended): Create a conda environment and install into it:
conda create -n directsketch sourmash_plugin_directsketch # create and install
conda activate directsketch # activate
Option 2: Install without creating an environment
conda install sourmash_plugin_directsketch
When you are downloading from a server, you may want to pay attention to the number of simultaneous downloads recommended/allowed. For NCBI (gbsketch, potentially urlsketch), the documentation is not currently clear on the number of simultaneous downloads allowed for fetch links. Here, I've limited to 30 (maximum, set via --n-simultaneous-downloads), which mirrors the limits set in the datasets tool provided by NCBI. Because of this, you may want to limit yourself to running just run a single instance of directsketch at once (e.g. with up to 30 simulataneous downloads), and/or enable batching for easier restart. Note that the number of simultaneous downloads impacts memory usage: see below for guidance.
If you're building large databases, we highly recommend you use batched zipfiles (v0.4+) to facilitate restart. If you encounter unexpected failures and are using a single zipfile output (default), gbsketch/urlsketch will have to re-download and re-sketch all files. If you instead set a batch size using --batch-size, then gbsketch/urlsketch can load any batched zips that finished writing, and avoid re-generating those signatures. The batch size represents the number of files downloaded, so it is possible DNA and protein signatures of the same accession may be split across zipfiles. Note that batches will use the --output file to build batched filenames, so if you provided output.sig.zip, batches will be output.1.sig.zip, etc (or output.zip --> output.N.zip). For small genomes (e.g. microbes), you can keep batch sizes quite large, e.g. 1000s-10000s. For large eukaryotic genomes where download takes much longer, you may want to use smaller batch sizes. At the end of a batched run, we write a file {output}.batchlist.txt with a list of all batch files created. This can then be used to build a single zipfile with sourmash sig cat (sourmash sig cat {output}.batchlist.txt -o OUTPUT.sig.zip) or sourmash sig collect to collect all the zips into a standalone manifest that can be used with sourmash and branchwater commands.
With batching, you can use the same exact command to restart: directsketch will detect already-generated signatures and downloaded FASTAs and skip them. It will start writing signature from the next batch index, so if you last completed zipfile was output.2.sig.zip, running the command again with the same --output output.sig.zip will start from output.3.sig.zip. Incomplete zipfiles (marked as .incomplete) will be deleted. The failure file will be overwritten, but since we retry any missing sketches, all failures will be re-attempted and then rewritten to the failure file if not successful. You can add the option --allow-completed to exit cleanly if there are existing signature batches present but no new signatures can be created/written. This will still write the failure file and {output}.batchlist.txt file, but no new zipfiles will be created.
Without batching, the existing signature zipfile will be unreadable and marked as .incomplete. You can restart by using the same input csv file, but if you are using --keep-fasta, you can add the option --no-overwrite-fasta to avoid overwriting any completed FASTA files. Incomplete fasta files are marked with the .incomplete extension and will be re-downloaded. If sketching, the incomplete zipfile will be deleted upon re-run, and we will still re-download all fasta files needed to generate sketches.
Directsketch v0.6+ streams the downloaded data, sketching and/or writing as it goes. For gzipped files, the library we use checks the internal crc32 to make sure we obtained the full download. urlsketch can also verify a user-provided md5sum. While you don't need to hold entire files in memory, you do need enough memory to hold chunks of downloaded data and signatures while sketching. You can limit the number of concurrent downloads (--n-simultaneous-downloads) to avoid memory issues. While testing with 10 eukaroyotic genomes under 1Gb each (10 simultaneous downloads), we used a maximum of ~2.5Gb.
gbsketch uses the NCBI REST API to download a dehydrated file with direct download links for the genomes. An API key may be needed if you run gbsketch many times, as NCBI has altered its download limitations. If you are unable to download the dehydrated zipfile, try providing an API Key. To obtain one, follow the instructions here. Once you have a key, you can provide it via the command line or set the NCBI_API_KEY variable (export NCBI_API_KEY=YOUR_KEY), which gbsketch will check and use automatically.
download and sketch NCBI Assembly Datasets by accession
First, create a file, e.g. acc.csv with GenBank identifiers and sketch names.
accession,name
GCA_000961135.2,GCA_000961135.2 Candidatus Aramenus sulfurataquae isolate AZ1-45
GCA_000175555.1,GCA_000175555.1 ACUK01000506.1 Saccharolobus solfataricus 98/2
Two columns must be present:
accession, andname.
To test gbsketch, you can download a csv file and run:
curl -JLO https://raw.githubusercontent.com/sourmash-bio/sourmash_plugin_directsketch/main/tests/test-data/acc.csv
sourmash scripts gbsketch acc.csv -o test-gbsketch.zip -f out_fastas -k --failed test.failed.csv --checksum-fail test.checksum-failed.csv -p dna,k=21,k=31,scaled=1000,abund -p protein,k=10,scaled=100,abund -r 1
To check that the zip was created properly, you can run:
sourmash sig summarize test-gbsketch.zip
and you should see the following as output:
** loading from 'test-gbsketch.zip'
path filetype: ZipFileLinearIndex
location: /path/to/your/test-gbsketch.zip
is database? yes
has manifest? yes
num signatures: 5
** examining manifest...
total hashes: 10815
summary of sketches:
2 sketches with dna, k=21, scaled=1000, abund 2884 total hashes
2 sketches with dna, k=31, scaled=1000, abund 2823 total hashes
1 sketches with protein, k=10, scaled=100, abund 5108 total hashes
Full Usage:
usage: gbsketch [-h] [-q] [-d] [-o OUTPUT] [-f FASTAS] [--batch-size BATCH_SIZE] [-k] [--download-only] [--failed FAILED] [--checksum-fail CHECKSUM_FAIL] [-p PARAM_STRING] [-c CORES] [-r RETRY_TIMES]
[-n {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30}] [-a API_KEY] [-v] [--write-urlsketch-csv] [--no-overwrite-fasta] [-g | -m]
input_csv
download and sketch GenBank assembly datasets
positional arguments:
input_csv A txt file or csv file containing accessions in the first column.
options:
-h, --help show this help message and exit
-q, --quiet suppress non-error output
-d, --debug provide debugging output
-o OUTPUT, --output OUTPUT
Output ZIP file for the signatures. Must end with '.zip'.
-f FASTAS, --fastas FASTAS
Write fastas here
--batch-size BATCH_SIZE
Write smaller zipfiles, each containing sigs associated with this number of accessions. This allows gbsketch to recover after unexpected failures, rather than needing to restart sketching from scratch. Default: write all
sigs to single zipfile.
-k, --keep-fasta Write FASTA files. Default: do not write FASTA files.
--download-only Download FASTAS but do not sketch. Requires '--keep-fasta'. By default this downloads both genomes and proteomes.
--failed FAILED CSV of failed accessions and download links (should be mostly protein).
--checksum-fail CHECKSUM_FAIL
CSV of accessions where the md5sum check failed or the md5sum file was improperly formatted or could not be downloaded.
-p PARAM_STRING, --param-string PARAM_STRING
Parameter string for sketching (default: k=31,scaled=1000).
-c CORES, --cores CORES
Number of cores to use (default is all available).
-r RETRY_TIMES, --retry-times RETRY_TIMES
Number of times to retry failed downloads (default=3).
-n {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30}, --n-simultaneous-downloads {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30}
Number of files to download simultaneously (1-30; default=10).
-a API_KEY, --api-key API_KEY
API Key for NCBI REST API. Alternatively, set NCBI_API_KEY environmental variable. If provided, will be used when downloading the initial dehyrated file.
-v, --verbose print progress for every download.
--write-urlsketch-csv
Write urlsketch-formatted csv with all direct download links. Will be '{input_csv}.urlsketch.csv'.
--no-overwrite-fasta Requires `--keep-fasta`. If set, do not overwrite existing FASTA files in the --fastas directory. Will still re-download those files if needed for sketching.
--allow-completed If batching and you've restarted but all feasible signatures have already been created (no new signatures can be created/written), exit cleanly anyway.
-g, --genomes-only Download and sketch genome (DNA) files only.
-m, --proteomes-only Download and sketch proteome (protein) files only.
download and sketch directly from URL(s)
First, create a file, e.g. acc-url.csv with identifiers, sketch names, and other required info.
accession,name,moltype,md5sum,download_filename,url,range
GCA_000961135.2,GCA_000961135.2 Candidatus Aramenus sulfurataquae isolate AZ1-454,dna,47b9fb20c51f0552b87db5d44d5d4566,GCA_000961135.2_genomic.urlsketch.fna.gz,https://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/961/135/GCA_000961135.2_ASM96113v2/GCA_000961135.2_ASM96113v2_genomic.fna.gz,
GCA_000961135.2,GCA_000961135.2 Candidatus Aramenus sulfurataquae isolate AZ1-454,protein,fb7920fb8f3cf5d6ab9b6b754a5976a4,GCA_000961135.2_protein.urlsketch.faa.gz,https://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/961/135/GCA_000961135.2_ASM96113v2/GCA_000961135.2_ASM96113v2_protein.faa.gz,
GCA_000175535.1,GCA_000175535.1 Chlamydia muridarum MopnTet14 (agent of mouse pneumonitis) strain=MopnTet14,dna,a1a8f1c6dc56999c73fe298871c963d1,GCA_000175535.1_genomic.urlsketch.fna.gz,https://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/175/535/GCA_000175535.1_ASM17553v1/GCA_000175535.1_ASM17553v1_genomic.fna.gz,
Six columns must be present:
accession- an accession or unique identifier. Ideally no spaces.name- full name for the sketch. Sourmash convention is to use '{accession} rest of name' as the signature name.moltype- is the file 'dna' or 'protein'?md5sum- expected md5sum(s). Optional, will be checked after download if provided.download_filename- filename for FASTA download. Required if--keep-fastas, but useful for signatures, too (saved in sig data).url- direct link(s) for the file(s)range- if desired, include base pair range(s), e.g 500-10000. This range will be selected from the record(s) and sketched (and/or saved to the download_filename). If there are multiple records in a FASTA file, the range will be applied to each record.
As of v0.5.0, urlsketch allows specification of multiple URLs to be downloaded and sketched into a single signature. If providing multiple URLs for a single accession/name, you must either provide no md5sum or range, or the number of entries in these columns must match the number of URLs. In each case, separate the entries with ';' -- e.g. "abc;def" for two md5sums.
To run after creating file above:
sourmash scripts urlsketch acc-url.csv -o test-urlsketch.zip -f out_fastas -k --failed test.failed.csv -p dna,k=21,k=31,scaled=1000,abund -p protein,k=10,scaled=100,abund -r 1
Full Usage:
usage: urlsketch [-h] [-q] [-d] [-o OUTPUT] [--batch-size BATCH_SIZE] [-f FASTAS] [-k] [--download-only] [--failed FAILED] [--checksum-fail CHECKSUM_FAIL] [-p PARAM_STRING] [-c CORES] [-r RETRY_TIMES]
[-n {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30}] [--force] [-v] [--no-overwrite-fasta] [-g | -m]
input_csv
download and sketch GenBank assembly datasets
positional arguments:
input_csv A txt file or csv file containing accessions in the first column.
options:
-h, --help show this help message and exit
-q, --quiet suppress non-error output
-d, --debug provide debugging output
-o OUTPUT, --output OUTPUT
Output ZIP file for the signatures. Must end with '.zip'.
--batch-size BATCH_SIZE
Write smaller zipfiles, each containing sigs associated with this number of urls. This allows urlsketch to recover after unexpected failures, rather than needing to restart sketching from scratch. Default: write all sigs
to single zipfile.
-f FASTAS, --fastas FASTAS
Write fastas here.
-k, --keep-fasta, --keep-fastq
Write FASTA/Q files. Default: do not write FASTA files.
--download-only Download FASTAS but do not sketch. Requires '--keep-fasta/--keep-fastq'.
--failed FAILED CSV of failed accessions and download links.
--checksum-fail CHECKSUM_FAIL
CSV of accessions where the md5sum check failed. If not provided, md5sum failures will be written to the download failures file (no additional md5sum information).
-p PARAM_STRING, --param-string PARAM_STRING
Parameter string for sketching (default: k=31,scaled=1000).
-c CORES, --cores CORES
Number of cores to use (default is all available).
-r RETRY_TIMES, --retry-times RETRY_TIMES
Number of times to retry failed downloads (default=3).
-n {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30}, --n-simultaneous-downloads {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30}
Number of files to download simultaneously (1-30; default=10).
--force Skip input rows with empty or improper URLs. Warning: these will NOT be added to the failures file.
-v, --verbose print progress for every download.
--no-overwrite-fasta Requires `--keep-fasta`. If set, do not overwrite existing FASTA files in the --fastas directory. Will still re-download those files if needed for sketching.
--allow-completed If batching and you've restarted but all feasible signatures have already been created (no new signatures can be created/written), exit cleanly anyway.
-g, --genomes-only Download and sketch genome (DNA) files only.
-m, --proteomes-only Download and sketch proteome (protein) files only.
This project is under the sourmash Code of Conduct.
We suggest filing issues in the directsketch issue tracker or the main sourmash issue tracker.
- N. Tessa Pierce-Ward
- C. Titus Brown
sourmash_plugin_directsketch is developed at https://github.com/sourmash-bio/sourmash_plugin_directsketch.
Run:
pytest tests
Bump version number in Cargo.toml and pyproject.toml and push.
Make a new release on github.
Then pull, and:
make sdist
Make sure directory is clean to avoid pulling in additional files.
followed by make upload_sdist.
you may need to
pip install twineif it is not available.