unipept / fraggenescanrs

Better and faster Rust implementation of the FragGeneScan gene prediction model for short and error-prone reads.

Unipept

The Unipept web application supports biodiversity and functional analysis of large and complex metaproteome samples and the analysis of peptidomes.

The 4.0 release of Unipept brings functional analysis to the tool.

An API and command line tool are available for integration in other programs.

Contributing

Found a bug or have an idea for an awesome new feature? File an issue using the github issue tracker or drop us a line at [email protected].

If you're willing to get your hands dirty, you might of course also send us a pull request!

Installation

This application is deployed and fully functional at unipept.ugent.be. If for some reason you wish to run your own instance, you can do so by deploying this rails application and setting up a database. This isn't straightforward and you'll probably want some help, so contact us at [email protected] before you attempt an installation.

Check our Wiki-pages for a variety of different installation guides.

Who made this app?

Unipept is a research project of the computational biology group at Ghent University. If you use this application, please cite:

• Gurdeep Singh et al. (2019) J. Proteome Res. doi:10.1021/acs.jproteome.8b00716
• Mesuere et al. (2018) J. Proteomics doi:10.1016/j.jprot.2017.05.022
• Mesuere et al. (2016) Proteomics doi:10.1002/pmic.201600023
• Mesuere et al. (2016) Bioinformatics doi:10.1093/bioinformatics/btw039
• Mesuere et al. (2015) Proteomics doi:10.1002/pmic.201400361
• Mesuere et al. (2012) J. Proteome Res. doi:10.1021/pr300576s

Current team:

• Bart Mesuere (@bmesuere): Postdoc and lead developer
• Pieter Verschaffelt (@pverscha): PhD student
• Tibo Vande Moortele (@tibvdm): PhD student
• Peter Dawyndt (@pdawyndt): Group leader and PhD supervisor

Other contributions from:

• Felix Van der Jeugt (@ninewise): Master's student 2014 - 2016 and PhD student 2016 - 2022
• Robbert Gurdeep Singh (@beardhatcode): Developer 2017-2018
• Tom Naessens (@silox): Master's student 2014-2015
• Toon Willems (@nudded): Master's student 2013-2014
• Ewan Higgs (@ehiggs): Ghent University HPC team
• Peter Vandamme: PhD co-supervisor of Bart
• Bart Devreese: PhD co-supervisor of Bart

For code contributions, the contributors graph is the place to be.

fraggenescanrs's Issues

gff format output

Hello FragGeneScanRs group,

I do not see gff format output as the original one does and also FGS+. This could be quite useful for quantification of reads mapped to genes in a lot of applications.

thanks,

Jianshu

Allow selection of translation table

Let the user choose a codon translation table instead of always using the 11th table. Would also require changing the start/stop-codon checks in the algorithm and possibly retraining the HMM parameters.

Add gzipped input support

Can you also add a support for gzipped fasta file as input? Should be very easy to implement? Or is is already supported? Think about when we have huge number of genomes downloaded from NCBI, which are all in *.fna.gz format.

(request by @jianshu93 in #5)

Better installation instructions in the readme

The installation instructions in the readme should be a bit longer and also targeted towards non-computer scientists. The recommended way is probably using cargo. The instructions to compile it from source are missing.

Add benchmark on assembly

The meta/evaluation directory contains a result quality benchmark on a complete genome. A benchmark on contigs should be added as well. First step would be to find a properly annotated assembly.

results not consistent with FGS+ and prodigal

Hello,

A benchmark test using FGS+ and prodigal showed that this FragGeneScanRs is not consistent with FGS+ and prodigal for most of the time (e.g. gene position predicted is very different) while FGS+ and prodigal are consistent with each other. I attache the genome I was using. There might be something wrong with the model or parameter setup.

.out is for rust version and .prodigal.gff is from prodigal

commands are:

FragGeneScanRs -s NTO_MAG_1.fasta -p 8 -t complete -w 1 -o NTO_MAG_1_gene

FGS+ -s NTO_MAG_1.fasta -p 8 -t complete -w 1 -o NTO_MAG_1_gene+

prodigal -i NTO_MAG_1.fasta -p meta -f gff -o NTO_MAG_1_prodigal.gff -a NTO_MAG_1.prodigal.faa -d NTO_MAG_1.prodigal.fna

NTO_MAG_1.fasta.zip

NTO_MAG_1_gene.out.zip
NTO_MAG_1_prodigal.gff.zip

Thanks,

Jianshu

Help command prints incorrect version number

Reported by a user: the version number is hard coded in https://github.com/unipept/FragGeneScanRs/blob/main/src/bin/FragGeneScanRs.rs#L31 and is outdated.

Add to Bioconda

FragGeneScanRs should be added to Bioconda to ease installation and use from Python. Integration in python could likely be similar to the existing FGS integration. Hurdle might be the rust compilation.

Error models for PacBio and Nanopore

First, love this tool.
Wondering if the error models for PacBio and Nanopore will be a including in a future release?
Or how would you recommend to include it?

Many thanks,
Rick

prodigal-rs

Hello Felix,

Hope you are doing well. A I mentioned earlier to you, prodigal, designed based on GeneMark, is widely used for complete microbial genomes gene prediction.