Visit our web site : http://frogs.toulouse.inrae.fr/
FROGS is a CLI workflow designed to produce an OTU count matrix from high depth sequencing amplicon data.
FROGS-wrappers allow to add FROGS on a Galaxy instance. (see https://github.com/geraldinepascal/FROGS-wrappers)
This workflow is focused on:
- User-friendliness with lots of rich graphic outputs and the integration in Galaxy thanks to FROGS-wrappers.
- Accuracy with a clustering without global similarity threshold, the management of separated PCRs in the chimera removal step, and the management of multi-affiliations.
- Dealing of non overlapping pair of sequences from long amplicon like ITS, or RPB2.
- Speed with fast algorithms parallelisation and easy to use.
- Scalability with algorithms designed to support the data growth.
- Convenient input data
- Installation
- Memory and parallelisation advices
- Download databanks
- Troubleshooting
- License
- Copyright
- Citation
- Contact
Legend for the next schemas:
.: Complete nucleic sequence
!: Region of interest
*: PCR primers
- Paired-end classical protocol: In the paired-end protocol R1 and R2 may share a nucleic region. For example the amplicons on 16S V3-V4 regions can have a length between 350 and 500nt, with 2*300pb sequencing the overlap is between 250nt and 100nt.
From: To:
rDNA .........!!!!!!................ ......!!!!!!!!!!!!!!!!!!!.....
Ampl ****!!!!!!**** ****!!!!!!!!!!!!!!!!!!!****
R1 -------------- --------------
R2 -------------- --------------
In any case, the maximum overlap between R1 and R2 can be the complete overlap.
The minimum authorized overlap between R1 and R2 is 10nt. With less, the overlap can be incorrect, it will be rejected or considered as non overlap reads.
- Single-end classical protocol:
rDNA .........!!!!!!................
Ampl ****!!!!!!****
Read --------------
- Custom protocol
rDNA .....!!!!!!!!!!!!!!............
Ampl ****!!!!!!****
Read --------------
The amplicons can have a high length variability such as ITS. The R1 and R2 can have different length.
This FROGS repository is for command line user. If you want to install FROGS on Galaxy, please refer to FROGS-wrappers.
FROGS is written in Python 3.7 (with external numpy and Scipy libraries) , uses also home-made scripts written in PERL5 and R 3.6.
FROGS relies on different specific tools for each of the analysis steps.
| FROGS Tools | Dependancy | version tested |
|---|---|---|
| Preprocess and Remove_chimera | vsearch | 2.17.0 |
| Preprocess | flash (optional) | 1.2.11 |
| Preprocess | cutadapt (need to be >=2.8) | 3.1 |
| Clustering | swarm (need to be >=2.1) | 3.0.0 |
| ITSx | ITSx | 1.1.2 |
| Affiliation_OTU | NCBI BLAST+ | 2.10.1 |
| Affiliation_OTU | RDP Classifier | 2.0.3 |
| Affiliation_OTU | EMBOSS needleall | 6.6.0 |
| Tree | MAFFT | 7.475 |
| Tree | Fasttree | 2.1.10 |
| Tree / FROGSSTAT | plotly, phangorn, rmarkdown, phyloseq, DESeq2, optparse, calibrate, formattable, DT | R 3.6.3 |
| FROGSSTAT | pandoc | 2.11.3 |
PEAR is one of the most effective software for read pairs merging, but as its license is not free for private use, we can not distribute it in FROGS. If you work in an academic lab on a private Galaxy server, or if you have paid your license you can use PEAR in FROGS preprocess. For that you need to:
- have PEAR in your PATH or in the FROGS libexec directory. We have tested PEAR 0.9.10 version (last version 0.9.11).
- use
--merge-software pearoption in the preprocess.py command line
FROGS is now available on bioconda (https://anaconda.org/bioconda/frogs).
- to create a specific environment for a specific FROGS version
conda env create --name [email protected] --file frogs-conda-requirements.yaml
# to use FROGS, first you need to activate your environment
conda activate [email protected]
To check your installation you can type:
cd <FROGS_PATH>/test
# when using conda FROGS_PATH=<conda_env_dir>/[email protected]/share/FROGS_3.2.3
sh test.sh <FROGS_PATH> <NB_CPU> <JAVA_MEM> <OUT_FOLDER>
"Bioinformatic" tools are performed on a small simulated dataset of one sample replicated three times. "Statistical" tools are performed on an extract of the published results of Chaillou et al, ISME 2014
This test executes the FROGS tools in command line mode. Example:
[user@computer:/home/frogs/FROGS/test/]$ sh test.sh ../ 1 2 res
Step preprocess : Flash mardi 10 novembre 2020, 10:56:56 (UTC+0100)
Step preprocess : Vsearch mardi 10 novembre 2020, 10:59:57 (UTC+0100)
Step clustering mardi 10 novembre 2020, 11:02:51 (UTC+0100)
Step remove_chimera mardi 10 novembre 2020, 11:08:31 (UTC+0100)
Step otu filters mardi 10 novembre 2020, 11:13:43 (UTC+0100)
Step ITSx mardi 10 novembre 2020, 11:14:00 (UTC+0100)
Step affiliation_OTU mardi 10 novembre 2020, 11:14:01 (UTC+0100)
Step affiliation_filter: masking mode mardi 10 novembre 2020, 11:14:53 (UTC+0100)
Step affiliation_filter: deleted mode mardi 10 novembre 2020, 11:14:54 (UTC+0100)
Step affiliation_postprocess mardi 10 novembre 2020, 11:14:54 (UTC+0100)
Step normalisation mardi 10 novembre 2020, 11:14:55 (UTC+0100)
Step clusters_stat mardi 10 novembre 2020, 11:14:55 (UTC+0100)
Step affiliations_stat mardi 10 novembre 2020, 11:14:58 (UTC+0100)
Step biom_to_tsv mardi 10 novembre 2020, 11:15:05 (UTC+0100)
Step biom_to_stdBiom mardi 10 novembre 2020, 11:15:06 (UTC+0100)
Step tsv_to_biom mardi 10 novembre 2020, 11:15:06 (UTC+0100)
Step tree mardi 10 novembre 2020, 11:15:06 (UTC+0100)
Step phyloseq_import_data mardi 10 novembre 2020, 11:16:36 (UTC+0100)
Step phyloseq_composition mardi 10 novembre 2020, 11:18:00 (UTC+0100)
Step phyloseq_alpha_diversity mardi 10 novembre 2020, 11:19:31 (UTC+0100)
Step phyloseq_beta_diversity mardi 10 novembre 2020, 11:20:19 (UTC+0100)
Step phyloseq_structure mardi 10 novembre 2020, 11:20:45 (UTC+0100)
Step phyloseq_clustering mardi 10 novembre 2020, 11:21:59 (UTC+0100)
Step phyloseq_manova mardi 10 novembre 2020, 11:22:20 (UTC+0100)
Step deseq2_preprocess mardi 10 novembre 2020, 11:22:42 (UTC+0100)
Step deseq2_visualisation mardi 10 novembre 2020, 11:23:29 (UTC+0100)
Completed with success
If you have more than one CPU, it is recommended to increase the number of CPUs used by tools. All the CPUs must be on the same computer/node.
| Tool | RAM per CPU | Minimal RAM | Configuration example |
|---|---|---|---|
| Preprocess | 8Gb | - | 12 CPUs and 96 GB |
| Clustering | - | 10 Gb | 16 CPUs and 60 GB |
| ITSx / Remove_Chimera | 3Gb | 5Gb | 12 CPUs and 36 GB |
| Affiliation_OTU | - | 20 Gb | 30 CPUs and 300 GB |
Reference database are needed to filter contaminants, assign taxonomy to each OTU or filter ambiguities for hyper variable amplicon length.
We propose some databanks, that you simply need to download and extract.
Please take time to read individual README.txt and LICENCE.txt files.
-
Assignation databank
these databanks are formatted for NCBI Blast+ and RDP Classifier
available databases : http://genoweb.toulouse.inra.fr/frogs_databanks/assignation
-
Contaminant databank
these banks are formatted for NCBI Blast+
http://genoweb.toulouse.inrae.fr/frogs_databanks/contaminants
-
Hyper variable in length amplicon databank
This is simply fasta file.
With some old versions of glibc the virtual memory used by CPU is multiplicative.
| Nb CPUs | expected RAM consumtion | observed RAM consumption |
|---|---|---|
| 1 | 1 Gb | 1Gb |
| 2 | 2 Gb | 2*2 Gb |
| 3 | 3 Gb | 3*3 Gb |
| 4 | 4 Gb | 4*4 Gb |
The parameters memory and CPU provided in examples take into account this problem.
GNU GPL v3
2020 INRAE
Depending on which type of amplicon you are working on (mergeable or unmergeable), please cite one of the two FROGS publications: