genomepy: genes and genomes at your fingertips

<!--- -->

genomepy is designed to provide a simple and straightforward way to download and use genomic data. This includes (1) searching available data, (2) showing the available metadata, (3) automatically downloading, preprocessing and matching data and (4) generating optional aligner indexes. All with sensible, yet controllable defaults. Currently, genomepy supports Ensembl, UCSC, NCBI and GENCODE.

Pssst, hey there! Is genomepy not doing what you want? Does it fail? Is it clunky? Is the documentation unclear? Have any other ideas on how to improve it? Don't be shy and let us know!

Table of Contents

1. Installation
2. Quick usage
3. Command line
   1. Search
   2. Inspect
   3. Install
   4. Plugins
4. Python
5. Frequently Asked Questions
6. Getting help
7. Contributing
8. Citation
9. License

Installation

genomepy requires Python 3.7+

You can install genomepy via bioconda, pip or git.

Bioconda

bash $ conda install -c conda-forge -c bioconda 'genomepy>=0.15'

Pip

bash $ pip install genomepy

With the Pip installation, you will have to install additional dependencies, and make them available in your PATH.

To read/write bgzipped genomes you will have to install pysam.

If you want to use gene annotation features, you will have to install the following utilities:

• genePredToBed
• genePredToGtf
• bedToGenePred
• gtfToGenePred
• gff3ToGenePred

You can find the binaries here.

Git

bash $ git clone https://github.com/vanheeringen-lab/genomepy.git $ conda env create -n genomepy -f genomepy/environment.yml $ conda activate genomepy $ pip install -e genomepy

Quick usage

1. Find your genome: $ genomepy search zebrafish

Console output: bash name provider accession tax_id annotation species other_info GRCz11 Ensembl GCA_000002035.4 7955 ✓ Danio rerio 2017-08-Ensembl/2018-04 ^ Use name for genomepy install

1. Install your genome (with gene annotation): $ genomepy install --annotation GRCz11 --provider ensembl

The default genomes directory: ~/.local/share/genomes/

Command line interface

All commands come with a short explanation when appended with -h/--help.

```bash $ genomepy --help Usage: genomepy [OPTIONS] COMMAND [ARGS]...

Options: --version Show the version and exit. -h, --help Show this message and exit.

Commands: annotation show 1st lines of each annotation clean remove provider data config manage configuration genomes list available genomes install install a genome & run active plugins plugin manage plugins providers list available providers search search for genomes ```

Search genomes & gene annotations

Let's say we want to download a Xenopus tropicalis genome & gene annotation. First, lets find out what's out there!

You can search by name, taxonomy ID or assembly accession ID. Additionally, you can limit the search result to one provider with -p/--provider. Furthermore, you can get the absolute --size of each genome (this option slows down the search).

bash $ genomepy search xenopus tro name provider accession tax_id annotation species other_info n r e k Xenopus_tropicalis_v9.1 Ensembl GCA_000004195.3 8364 ✓ Xenopus tropicalis 2019-04-Ensembl/2019-12 xenTro1 UCSC na 8364 ✗ ✗ ✗ ✗ Xenopus tropicalis Oct. 2004 (JGI 3.0/xenTro1) xenTro2 UCSC na 8364 ✗ ✓ ✓ ✗ Xenopus tropicalis Aug. 2005 (JGI 4.1/xenTro2) xenTro3 UCSC GCA_000004195.1 8364 ✗ ✓ ✓ ✗ Xenopus tropicalis Nov. 2009 (JGI 4.2/xenTro3) xenTro7 UCSC GCA_000004195.2 8364 ✓ ✓ ✗ ✗ Xenopus tropicalis Sep. 2012 (JGI 7.0/xenTro7) xenTro9 UCSC GCA_000004195.3 8364 ✓ ✓ ✓ ✗ Xenopus tropicalis Jul. 2016 (Xenopus_tropicalis_v9.1/xenTro9) Xtropicalis_v7 NCBI GCF_000004195.2 8364 ✓ Xenopus tropicalis DOE Joint Genome Institute Xenopus_tropicalis_v9.1 NCBI GCF_000004195.3 8364 ✓ Xenopus tropicalis DOE Joint Genome Institute UCB_Xtro_10.0 NCBI GCF_000004195.4 8364 ✓ Xenopus tropicalis University of California, Berkeley ASM1336827v1 NCBI GCA_013368275.1 8364 ✗ Xenopus tropicalis Southern University of Science and Technology ^ Use name for genomepy install

Inspect gene annotations

Let's say we want to download the Xenopus tropicalis genome & gene annotation from UCSC.

Since we are interested in the gene annotation as well, we should check which gene annotation suits our needs. As you can see in the search results, UCSC has several gene annotations for us to choose from. In the search results, n r e k denotes which UCSC annotations are available. These stand for ncbiRefSeq, refGene, ensGene and knownGene, respectively.

We can quickly inspect these with the genomepy annotation command:

bash $ genomepy annotation xenTro9 -p ucsc 12:04:41 | INFO | UCSC ncbiRefSeq chr1 genomepy transcript 133270 152620 . - . gene_id "LOC100490505"; transcript_id "XM_012956089.1"; gene_name "LOC100490505"; chr1 genomepy exon 133270 134186 . - . gene_id "LOC100490505"; transcript_id "XM_012956089.1"; exon_number "1"; exon_id "XM_012956089.1.1"; gene_name "LOC100490505"; 12:04:45 | INFO | UCSC refGene chr1 genomepy transcript 193109390 193134311 . + . gene_id "pias2"; transcript_id "NM_001078987"; gene_name "pias2"; chr1 genomepy exon 193109390 193109458 . + . gene_id "pias2"; transcript_id "NM_001078987"; exon_number "1"; exon_id "NM_001078987.1"; gene_name "pias2"; 12:04:49 | INFO | UCSC ensGene chr1 genomepy transcript 133270 152620 . - . gene_id "ENSXETG00000030302.2"; transcript_id "ENSXETT00000061673.2"; gene_name "ENSXETG00000030302.2"; chr1 genomepy exon 133270 134186 . - . gene_id "ENSXETG00000030302.2"; transcript_id "ENSXETT00000061673.2"; exon_number "1"; exon_id "ENSXETT00000061673.2.1"; gene_name "ENSXETG00000030302.2";

Here we can see that the refGene annotation has actual HGNC gene names, so lets go with this annotation. This differs between assemblies, so be sure to check!

Install a genome & gene annotation

Copy the name returned by the search function to install.

bash $ genomepy install xenTro9

You can choose to download gene annotation files with the -a/--annotation option.

bash $ genomepy install xenTro9 --annotation

For UCSC we can also select the annotation type. See genomepy install --help for all provider specific options.

bash $ genomepy install xenTro9 --UCSC-annotation refGene

Since we did not specify the provider here, genomepy will use the first provider with xenTro9. You can specify a provider by name with -p/--provider:

```bash $ genomepy install xenTro9 -p UCSC Downloading genome from http://hgdownload.soe.ucsc.edu/goldenPath/xenTro9/bigZips/xenTro9.fa.gz... Genome download successful, starting post processing...

name: xenTro9 local name: xenTro9 fasta: ~/.local/share/genomes/xenTro9/xenTro9.fa ```

Next, the genome is downloaded to the directory specified in the config file (by default ~/.local/share/genomes). To choose a different directory, use the -g/--genomes_dir option:

```bash $ genomepy install sacCer3 -p UCSC -g /path/to/my/genomes Downloading genome from http://hgdownload.soe.ucsc.edu/goldenPath/sacCer3/bigZips/chromFa.tar.gz... Genome download successful, starting post processing...

name: sacCer3 local name: sacCer3 fasta: /path/to/my/genomes/sacCer3/sacCer3.fa ```

Regex, masking & compression

You can use a regular expression to filter for matching sequences (or non-matching sequences by using the -n/--no-match option). For instance, the following command downloads hg38 and saves only the major chromosomes:

```bash $ genomepy install hg38 -p UCSC -r 'chr[0-9XY]+$' Downloading genome from from http://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/hg38.fa.gz... Genome download successful, starting post processing...

name: hg38 local name: hg38 fasta: /data/genomes/hg38/hg38.fa

$ grep ">" /data/genomes/hg38/hg38.fa

chr1 chr10 chr11 chr12 chr13 chr14 chr15 chr16 chr17 chr18 chr19 chr2 chr20 chr21 chr22 chr3 chr4 chr5 chr6 chr7 chr8 chr9 chrX chrY ```

By default, genome sequences are soft-masked (ACgtN). Use -m hard for hard masking (ACNNN), or -m none for no masking (ACGTN).

bash $ genomepy install hg38 --mask hard

If you wish to conserve space, you can tell genomepy to compress the downloaded data by passing the -b/--bgzip option. See Configuration for details.

bash $ genomepy install hg38 --bgzip

Other providers (any URL/local files)

To use assemblies not on NCBI, UCSC, Ensembl or GENCODE, you can give a URL instead of a name, together with --provider URL. Similarly, if you have a local FASTA file, you can install this using the filepath, together with --provider Local:

bash $ genomepy install -p url https://research.nhgri.nih.gov/hydra/download/assembly/\Hm105_Dovetail_Assembly_1.0.fa.gz

This will install the genome under the filename of the URL/filepath, but can be changed with the -l/--localname option.

If you add the --annotation flag, genomepy will search the (remote) directory for an annotation file as well. Should this fail, you can also add a URL to the annotation with --URL-to-annotation with the URL provider, or a filepath with --Local-path-to-annotation with the Local provider:

bash $ genomepy install -p local /path/to/genome.fa --Local-path-to-annotation /path/to/gene.annotation.gtf

Reproducibility

All selected options are stored in a README.txt. This includes the original name, download location and other genomepy operations (such as regex filtering and time).

Plugins

Plugins are optional steps that are executed after installing an assembly with genomepy install. If you already installed an assembly, you can activate a plugin and rerun the install command. This will not overwrite your local files, unless you use the --force option.

Check which plugins are enabled with genomepy plugin list.

Genome blacklists

For some model organisms, genomepy can download a genome blacklist (generated by the Kundaje lab). Blacklists are only available for these model organisms when downloaded from UCSC, and for the human and mouse genomes.

Enable the blacklist plugin to use it:

bash $ genomepy plugin enable blacklist Enabled plugins: blacklist

Aligner indexes

You can also create aligner indexes for several widely used aligners. Currently, genomepy supports:

• Bowtie2
• BWA
• GMAP
• HISAT2
• Minimap2
• STAR

These programs are not installed by genomepy and need to be installed separately for the indexing to work. The easiest way to do so is with conda, e.g.: conda install -c bioconda bwa star

Splice-aware indexing (required for e.g. RNA-seq) can be performed by STAR and Hisat2. This will be done automatically if the gene annotation was downloaded as well. Finally, STAR can further improve mapping to (novel) splice junctions by indexing again (see 2-pass mapping mode in the STAR manual). The second pass is not supported by genomepy.

You can configure the index creation with genomepy plugin enable, e.g.:

bash $ genomepy plugin enable bwa star Enabled plugins: blacklist, bwa, star

You can pass the number of threads to use for aligner index creation with genomepy install --threads (default is 8).

Configuration

All defaults can be overwritten on the command line and in Python. However, you can create & edit the config file to change the default settings (full description):

bash $ genomepy config generate Created config file ~/.config/genomepy/genomepy.yaml

Genome location

By default, genomes will be saved in ~/.local/share/genomes.

To set the default genome directory, to /data/genomes for instance, edit ~/.config/genomepy/genomepy.yaml and change the following line:

yaml genomes_dir: /data/genomes

Compression

Genome FASTA files can be stored using bgzip compression. This means that the FASTA files will take up less space on disk. Set the following line to your config file:

yaml bgzip: True

Most tools are able to use bgzip-compressed genome files. One notable exception is bedtools getfasta. As an alternative, you can use the faidx command-line script from pyfaidx which comes installed with genomepy.

List available providers

bash $ genomepy providers GENCODE Ensembl UCSC NCBI Local URL

List available genomes

You can constrain the genome list by using the -p/--provider option to search only a specific provider. Additionally, you can get the absolute --size of each genome (this option slows down the search).

bash $ genomepy genomes -p UCSC name provider accession tax_id annotation species other_info n r e k ailMel1 UCSC GCF_000004335.2 9646 ✓ ✗ ✓ ✗ Ailuropoda melanoleuca Dec. 2009 (BGI-Shenzhen 1.0/ailMel1) allMis1 UCSC GCA_000281125.1 8496 ✗ ✓ ✗ ✗ Alligator mississippiensis Aug. 2012 (allMis0.2/allMis1) anoCar1 UCSC na 28377 ✗ ✗ ✓ ✗ Anolis carolinensis Feb. 2007 (Broad/anoCar1)

Local cache.

Note that the first time you run genomepy search or list the command will take a while as the genome lists have to be downloaded. The lists are cached locally, which will save time later. The cached files are stored in ~/.cache/genomepy and expire after 7 days (so they stay up to date). This expiration time can be changed in the config file. You can also delete this directory to clean the cache using genomepy clean.

Python API

Check out our Python API documentation here

```

import genomepy for row in genomepy.search("GRCh38"): ... print(row) ...
['GRCh38.p14', 'Ensembl', 'GCA000001405.28', 9606, True, 'Homo sapiens', '2014-01-Ensembl/2021-03'] ['hg38', 'UCSC', 'GCA000001405.15', 9606, [True, True, False, True], 'Homo sapiens', 'Dec. 2013 (GRCh38/hg38)'] ['GRCh38', 'NCBI', 'GCF000001405.26', 9606, True, 'Homo sapiens', 'Genome Reference Consortium'] ['GRCh38.p1', 'NCBI', 'GCF000001405.27', 9606, True, 'Homo sapiens', 'Genome Reference Consortium'] ['GRCh38.p2', 'NCBI', 'GCF000001405.28', 9606, True, 'Homo sapiens', 'Genome Reference Consortium'] ['GRCh38.p3', 'NCBI', 'GCF000001405.29', 9606, True, 'Homo sapiens', 'Genome Reference Consortium']

genomepy.installgenome("hg38", annotation=True, provider="UCSC", genomesdir="./data/genomes") Downloading genome from UCSC. Target URL: http://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/hg38.fa.gz... Genome download successful, starting post processing... name: hg38 local name: hg38 fasta: ./data/genomes/hg38/hg38.fa Downloading the ncbiRefSeq annotation from the UCSC MySQL database. Annotation download successful

a = genomepy.Annotation("hg38", genomesdir="./data/genomes") a.namedgtf.head(3) seqname ... attribute genename ...
DDX11L1 chr1 ... geneid "DDX11L1"; transcriptid "NR046018.2"... DDX11L1 chr1 ... geneid "DDX11L1"; transcriptid "NR046018.2"... DDX11L1 chr1 ... geneid "DDX11L1"; transcriptid "NR046018.2"...

start = a.namedgtf.loc["TP63"]["start"].min() end = a.namedgtf.loc["TP63"]["end"].max() chrom = a.named_gtf.loc["TP63"]["seqname"][0]

g = genomepy.Genome("hg38", genomes_dir="./data/genomes") g[chrom][start:end] chr3:189596747-189897276 gcaacccgctggggtcaccttccacactgtggaagctttgttcttttgctctttgcagtaaatcttgct...

```

The genomepy.Genome class builds on top of the pyfaidx.Fasta class, see the pyfaidx documentation for more details. The genomepy.Annotation class contains pandas Dataframes with GTF and BED files, as well as additional class methods to utilize these.

Frequently Asked Questions

Genomepy utilizes external databases to obtain your files. Unfortunately this sometimes causes issues. Here are some of the more common issues, with solutions.

Let us know if you encounter issues you cannot solve by creating a new issue.

Provider is offline

Occasionally one of the providers experience connection issues, which can last anywhere between minutes to hours. When this happens genomepy will warn that the provider appears offline, or that the URL seems broken.

If the issue does not pass, you can try to reset genomepy. Simply run genomepy clean on the command line, or run genomepy.clean() in Python.

This genome is missing

Genomepy stores provider data on your computer to rerun it faster later. If a provider was offline during this time, it may miss (parts of) the data.

To re-download the data, remove the local data with genomepy clean, then search for your genome again.

This genome is STILL missing/URL is broken

Sadly, not everything (naming, structure, filenames) is always consistent on the provider end. Contact the provider to get it fixed! One notable group are Ensembl fungi, which seems to be mostly mislabelled.

In the meantime, you can still use the power of genomepy by manually retrieving the URLs, and downloading the files with genomepy install GENOME_URL -p url --url-to-annotation ANNOTATION_URL.

The genomepy config was corrupted

You can create a new one with genomepy config generate on command line, or genomepy.manage_config("generate") in Python.

What's genomepy maximum memory usage?

Genomepy does not read a genome fully into memory. Therefore, installing takes less than 1 GB RAM regardless of the genome's size. Searching NCBI is the most costly operation, using around 3 GB (the first time).

Which genome/gene annotation to use

Each provider has its pros and cons: * Ensembl has excellent gene annotations, but their chromosome names can cause issues with some tools. * UCSC has an excellent genome browser, but their gene annotations vary in format. * NCBI allows public submissions, and so has the latest versions, although not always complete or error free.

Use genomepy search to see your options, and genomepy annotation to check the quality of the gene annotation(s).

Getting help

If you want to report a bug or issue, or have problems with installing or running the software please create a new issue. This is the preferred way of getting support. Alternatively, you can mail me.

Contributing

Contributions welcome! Send me a pull request or get in touch.

When contributing a PR, please use the develop branch.

Quick development setup:

1. Fork & download this repo.
2. cd into your local repo.
3. git checkout develop
4. conda env create -f environment.yaml
5. conda activate genomepy
6. pip install -e .
7. git checkout -b yourdevelopbranch

The command line and python imports will now use the code in your local repo. To test your changes locally, run the following command: pytest -vvv

Contributors

• Siebren Frölich - @siebrenf
• Maarten van der Sande - @Maarten-vd-Sande
• Tilman Schäfers @tilschaef
• Simon van Heeringen - @simonvh
• Dohoon Lee - @dohlee
• Jie Zhu - @alienzj

Citation

If you use genomepy in your research, please cite it: 10.1093/bioinformatics/btad119.

License

This module is licensed under the terms of the MIT license.