A Linux command-line tool to extract proteins from proteomes.
ProtExtract uses DIAMOND to search for potential homologs of proteins within proteomes and extracts the sequence with the highest bit-score sum among all alignments with sequences from a query protein file. It then performs a reciprocal DIAMOND search for verification, only keeping an extracted sequence if it realigns with a sequence from the same query protein file as before. The extracted and verified sequences will be annotated with the filename of the query protein. This tool was primarily designed to work with the 320 marker genes dataset from Strassert et al. 2021. While it should work with other single-copy protein datasets, the results might be less accurate as the reciprocal verification depends on rich datasets. It only extracts the best hit for each query protein file to avoid extracting paralogs. If no homolog exists in the proteome, the best hit may be a paralog or analog. Various information files about the found and missing proteins are created as well, including an output table.
A directory with protein files as queries. The highest accuracy is achieved with large datasets comprising many protein files, each containing homologous sequences from numerous species.
- The protein filenames will be treated as the protein names and must not contain spaces.
- If any gaps exist within the protein sequences, they will be copied for gap removal.
- The default directory is
Proteinsand the default file extension is.fasta. - A custom directory can be provided with the option
-qand a custom file extension with-Q.
A directory with proteome files as subjects. Can be the same input folder, as long as the file extensions are different.
- The sequence headers should ideally not contain vertical bars
|. - The default directory is
Proteomesand the default file extension is.fasta. - A custom directory can be provided with the option
-sand a custom file extension with-S.
Linux or WSL 2 on Windows (might also work on MacOS).
ProtExtract_scripts folder
and its content placed in the same directory as the ProtExtract main script.
Conda environment called protextract with DIAMOND, Python 3, Biopython and natsort installed.
- Create this conda environment automatically using the
-c(with conda) or-C(with mamba) option.
To use ProtExtract, run the following command in the terminal:
./ProtExtract [options]
-q DIRECTORY: Protein (query) input directory. Defaults to 'Proteins'.
Add proteins for which homologs are to be extracted here. Cannot be symlinks.
-Q STRING: Protein (query) file extension. Defaults to '.fasta'.
-s DIRECTORY: Proteome (subject) input directory. Defaults to 'Proteomes'.
Add proteomes from which proteins are to be extracted here. Cannot be symlinks.
-S STRING: Proteome (subject) file extension. Defaults to '.fasta'.
-o DIRECTORY: Output directory. Defaults to 'Output_<evalue>_<coverage>_<pidentity>'.
-O STRING: Output directory suffix, appended with an underscore. Defaults to no suffix.
-t INTEGER: Number of threads used by DIAMOND. Defaults to all available cores.
-m STRING: Sensitivity mode for DIAMOND. Defaults to 'default'.
Available modes: 'fast', 'default', 'mid-sensitive', 'sensitive', 'more-sensitive', 'very-sensitive', 'ultra-sensitive'
-E STRING: Maximum allowed E-value for DIAMOND. Defaults to '1e-20'.
-% INTEGER: Minimum required percentage of query and subject coverage for DIAMOND. Defaults to '50'.
-p INTEGER: Minimum required percentage of identical positions for DIAMOND. Defaults to '20'.
-f: Force overwrite of existing output files.
-c: Create the required conda environment with Anaconda:
conda create -n protextract -c conda-forge -c bioconda python biopython natsort diamond
-C: Create the required conda environment with Mamba:
mamba create -n protextract -c conda-forge -c bioconda python biopython natsort diamond
-v: Display the version number.
-h: Show the help message.
To install the required conda environment, navigate to the directory containing the ProtExtract main script, enter the following command and follow the instructions:
./ProtExtract -c
To run ProtExtract with default options, navigate to the directory containing the ProtExtract main script and enter the following command:
./ProtExtract
This will run ProtExtract with the default options:
Proteins (queries): ./Proteins/*.fasta
Proteomes (subjects): ./Proteomes/*.fasta
Output directory: ./Output_1e-20_cov50_pid20/
Number of threads: all available cores
Sensitivity mode: default
Maximum allowed E-value: 1e-20
Minimum required coverage: 50%
Minimum required percent identity: 20%
To run ProtExtract with custom options, navigate to the directory containing the ProtExtract main script and enter the following command:
./ProtExtract -q ./my_proteins/ -Q .fa -s ./my_proteomes/ -S .pep -o ./my_output/ -O suffix -t 4 -m fast -E 1e-30 -% 60 -p 50
This will run ProtExtract with the following options:
Proteins (queries): ./my_proteins/*.fa
Proteomes (subjects): ./my_proteomes/*.pep
Output directory: ./my_output_suffix/
Number of threads: 4
Sensitivity mode: fast
Maximum allowed E-value: 1e-30
Minimum required coverage: 60%
Minimum required percent identity: 50%
0.1.0: First version. Homologous sequences are identified by lowest E-value.
0.1.1: Homologous sequences are now identified by highest bit-score, as multiple alignments with an E-value of 0 might exist.
0.2.0: Added options to provide the input proteins and proteomes, as well as options to show the help menu and the current version.
0.2.1: Homologous sequences are now identified by highest bit-score sum among all alignments with homologous sequences of a query protein.
0.3.0: Added an option to specify an output folder, including an overwrite check and a force overwrite option.
0.3.1: Added E-value, coverage, and percentage identity threshold options for DIAMOND.
0.4.0: Added options for automatic setup of the required Anaconda environment with Anaconda or Mamba.
0.4.1: Added query and subject coverage to 12th and 13th column of the output format 6.
0.5.0: Major feature update: Added reciprocal verification with DIAMOND.
0.5.1: Fixed a bug which sometimes caused unverified proteins to be kept as final homologs.
0.6.0: Gaps within query protein sequences will now be removed to improve DIAMOND results.
0.6.1: Switched to natural sorting of protein names wherever possible.
0.7.0: A protein summary table containing information about all found and missing proteins in the proteomes is now generated.
0.7.1: Fixed a bug which caused ProtExtract to crash when no proteins were found at all.
0.8.0: Added an option to select the sensitivity level of DIAMOND.
0.8.1: Runtime is now displayed at the end.
0.9.0: Cleaned up the code. First public version.
0.9.1: A single temporary file is now created to store all best hits instead of a separate file for each hit.
The default output directory name now contains the search options.
The presence of the necessary Python scripts is now verified before execution.
The Python scripts are now executed relative to the location of the main script.
Restructured the code slightly.
Specified use case in README.md.
1.0.0: Added an option to set the number of threads used by DIAMOND, defaulting to all available cores.
Added an option to set an output directory suffix, defaulting to no suffix.
The default sensitivity for DIAMOND is now 'default', which was not available before.
The number of extracted and verified proteins per proteome is now shown in the terminal output.
Renamed the folder 'ProtExtract_files' to 'ProtExtract_scripts'.
Improved log file handling.
Improved gap detection.
Updated README.md.
Justin Teixeira Pereira Bassiaridis
Distributed under the MIT License. See LICENSE for more information.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent