RRNPP_detector

1. What does RRNPP_detector do?

RRNPP_detector aims at identifying known and novel RRNPP-type quorum sensing systems in chromosomes, mobile genetic elements and bacteriophages of Firmicutes.

RRNPP_detector defines candidate RRNPP-type quorum sensing systems as tandems of adjacent ORFs encoding a candidate receptor (250-500aa protein matching HMMs of peptide-binding tetraticopeptide repeats (TPRs)) and a candidate pro-peptide (10-100aa protein predicted to be excreted via the SEC-translocon or matching the amino-acid profile of SHPs)

RRNPP_detector outputs candidate RRNPP-type systems with 3 different detection strictness:

• strict: All receptor-propeptide pairs oriented in a divergent or co-directional context with the propeptide downstream from the receptor, preceded by a RBS_bin > 13 and matching the HMM profile of SHPs or predicted to undergo a SEC/SPI-dependent secretion according to SignalP
• relaxed: All remaining receptor-propeptide pairs in which the propeptide harbors any of the SP(Sec/SPI), TAT(Tat/SPI) or LIPO(Sec/SPII) signal sequence according to PrediSi or SignalP
• loose: All remaining receptor-propeptide pairs in which the propeptide is preceded by a RBS motif with both a high score (according to Prodigal) and a high usage across prokaryotes (according to Omotajo et al.)

2. How to install RRNPP_detector?

Create a dedicated and isolated conda environment for RRNPP_detector (recommended but not mandatory)

conda create --name rrnpp_detector
conda activate rrnpp_detector

Install dependencies

Using conda

conda install -c bioconda pandas orfipy prodigal hmmer blast openjdk

orfipy needs sqlite3 to be compiled. If the installation exits with the "sqlite3.h: No such file or directory" error message, please try the following:

sudo apt-get install libsqlite3-dev
conda install -c bioconda pandas orfipy prodigal hmmer blast openjdk

Alternative to installation via conda:

sudo apt-get install prodigal hmmer ncbi-blast+
pip3 install pandas orfipy

Clone the repository

cd ~
git clone https://github.com/TeamAIRE/RRNPP_detector.git
cd RRNPP_detector/

Uncompress SignalP and compile PrediSi

tar -xvzf signalp-5.0b.tar.gz

# optional: if you intend to use Predisi in addition to SignalP
cd predisi
javac JSPP.java
cd ../

RRNPP_detector comes by default with the binary of SignalP version 5.0b Linux x86_64. However, if you want to use another version of SignalP, the software is freely available for academic users at: https://services.healthtech.dtu.dk/software.php

3. How to use RRNPP_detector?

General usage

RRNPP_detector must be run with python3 against one or multiple contig(s)/(meta)genome(s)

usage: rrnpp_detector.py [-h] [--version] [-o OUT_DIR] [--fna FNA] [--faa FAA] [--gff GFF] [--ft FEATURE_TBL] [--cpu CPU] [--preserve_ram] [--keep_working_dir] [--min_pl MIN_PROPEPTIDE_LEN]
                         [--max_pl MAX_PROPEPTIDE_LEN] [--min_rl MIN_RECEPTOR_LEN] [--max_rl MAX_RECEPTOR_LEN] [--min_igd MIN_INTERGEN_DIST] [--max_igd MAX_INTERGEN_DIST] [--expand_to_homologs]
                         [--tprpred] [--predisi]

RRNPP_detector: a tool to detect RRNPP-Type quorum sensing systems in chromosomes, plasmids and bacteriophages of Firmicutes

Main arguments
 -h, --help               show this help message and exit
  --version               print version number and exit.
  -o OUT_DIR              path to output directory (default is current directory)
  --fna FNA               path to the fasta of the target genome(s) (will run Prodigal to detect CDSs if faa not provided)
  --faa FAA               path to fasta of the protein sequences of the target genome(s) (requires additional --gff or --ft option)
  --gff GFF               path to the annotations of the target genome(s) in gff
  --ft FEATURE_TBL        path to the annotations of the target genome(s) in the NCBI_assembly feature_table format
  --cpu CPU               number of cpu to use (default is 1)
  --chunk_size CHUNK_SIZE nb target genomes to be processed altogether to preserve RAM usage 
                          (e.g. if --chunk_size 100, then the program will process 100 genomes by 100 genomes instead of all together)
  --keep_working_dir      keep the directory of intermediate files

Specific optional parameters for searching RRNPP systems
  --min_pl  MIN_PROPEPTIDE_LEN  minimal propeptide length (default=10)
  --max_pl  MAX_PROPEPTIDE_LEN  maximal propeptide length (default=70)
  --min_rl  MIN_RECEPTOR_LEN    minimal receptor length (default=250)
  --max_rl  MAX_RECEPTOR_LEN    maximal receptor length (default=500)
  --min_igd MIN_INTERGEN_DIST   minimal intergenic distance (default=-20)
  --max_igd MAX_INTERGEN_DIST   maximal intergenic distance (default=400)
  --start_codons START_CODONS   comma-separated list of start codons to consider for ORF calling (default=ATG)
  --rbs_bins RBS_BINS           comma-separated list of Prodigal's RBS bins to consider for ORF calling (default=27,24,23,22,20,19,16,15,14,13,12,6), 
                                to by bypass the filter, use --rbs_bins 27,26,25,24,23,22,21,20,19,18,17,16,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0
  --expand_to_homologs          use detected systems as seeds to detect putative homologous systems missed by RRNPP_detector
  --tprpred                     run tprpred in addition to hmmsearch for TPR motifs detection
  --predisi                     run predisi in addition to signalp for detection of propeptides with a signal sequence (warning: this increases the risk of false positives)

Best option: Provide one or multiple genome(s) (fna) along with annotation(s) (gff) and proteome(s) (faa)

When rrnpp_detector is fed with one or multiple genome(s) (concatenated in one fna file) along with their corresponding annotated proteome(s) (concatenated in one faa file) and annotations (concatenated in one gff or one NCBI_assembly feature_table file), both annotated proteins and unannotated proteins (searched in the genomic vicinity of annotated receptors) will be present in the output.

python rrnpp_detector.py --fna bacillus_subtilis/genome.fna --faa bacillus_subtilis/proteome.faa --gff bacillus_subtilis/annotations.gff

Alternative 1: Provide one or multiple genome(s) (fna)

If you provide only a nucleotide fasta file (fna) to rrnpp_detector, the tool won't work with already annotated proteins. Instead, the proteins will be detected with prodigal. Search for unannotated small ORFs in the vicinity of receptors will be performed.

python rrnpp_detector.py --fna bacillus_subtilis/genome.fna

Alternative 2: Provide all proteins from one or multiple genome(s) (faa) + the annotations of the genome(s) (gff)

If you don't provide a nucleotide fasta file but only a proteome file (faa) and an annotation file, RRNPP_detector will work only with annotated proteins. The downside of this option is that it won't search for unannotated small ORFs in the vicinity of receptors

# with a gff
python rrnpp_detector.py --faa bacillus_subtilis/proteome.faa --gff bacillus_subtilis/annotations.gff
# with a feature table
python rrnpp_detector.py --faa bacillus_subtilis/proteome.faa --ft bacillus_subtilis/feature_table.txt

4. Custom search

If you wish to design a custom search with very specific parameters, you can change the parameters hard coded in rrnpp_detector/preprocessing.py

5. Practical example of analysis

In this example, we will propose to use RRNPP_detector against all genomes of Viruses available on the NCBI.

First, we will use the ncbi_datasets command line tool to fetch the genomes from the NCBI assembly database. Here is a tutorial on how to install ncbi_datasets: https://www.ncbi.nlm.nih.gov/datasets/docs/v2/download-and-install/

conda activate ncbi_datasets

DOWNLOAD_DIRECTORY="/home/viral_genomes";
mkdir "$DOWNLOAD_DIRECTORY";

# will donwload genomes (fna), protein sequences (faa) and annotations (gff3) of complete genomes of Viruses available on Genbank
datasets download genome taxon "Viruses" \
--annotated \
--assembly-level complete \
--assembly-source genbank \
--include genome,protein,gff3 \
--filename "$DOWNLOAD_DIRECTORY"/assemblies.zip

Next, we will extract the files from this archive and concatenate all genomes in one genomes.fna file, all protein fastas in one proteomes.faa file and all gffs in one annotations.gff file

cd "$DOWNLOAD_DIRECTORY"
unzip assemblies.zip
cd ncbi_dataset/data
for ASSEMBLY in GCA_*; do 
  cat "$ASSEMBLY"/*.fna >> ../../viral_genomes.fna; rm "$ASSEMBLY"/*.fna;
  cat "$ASSEMBLY"/*.faa >> ../../viral_proteomes.faa; rm "$ASSEMBLY"/*.faa;
  cat "$ASSEMBLY"/*.gff >> ../../viral_annotations.gff; rm "$ASSEMBLY"/*.gff;
done

All that remains to do is to execute RRNPP_detector against this dataset. For instance:

cd ~/Programs/rrnpp_detector
python rrnpp_detector.py --fna "$DOWNLOAD_DIRECTORY"/viral_genomes.fna --faa "$DOWNLOAD_DIRECTORY"/viral_proteomes.faa --gff "$DOWNLOAD_DIRECTORY"/viral_annotations.gff -o ~/rrnpp_detector_vs_viruses --cpu 20

6. Help improving RRNPP_detector

If you have suggestions to improve the tool or would like to report bugs, please post your message on the Issues section of this repository.

7. Citation.

If you are using RRNPP_detector, please cite the following paper:

• Bernard C, Li Y, Lopez P, Bapteste E: Large-scale identification of known and novel RRNPP quorum sensing systems by RRNPP_detector captures novel features of bacterial, plasmidic and viral co-evolution. Molecular Biology and Evolution. 2023. doi: 10.1093/molbev/msad062

We strongly encourage you to cite the following 3rd party dependencies:

• Almagro Armenteros JJ, Tsirigos KD, Sønderby CK, Petersen TN, Winther O, Brunak S, von Heijne G, Nielsen H. 2019. Signalp 5.0 improves signal peptide predictions using deep neural networks. Nat Biotechnol. 37:420–423.
• Eddy SR. 2011. Accelerated profile HMM searches. PLoS Comput Biol. 7:e1002195.
• Hyatt D, Chen GL, LoCascio PF, Land ML, Larimer FW, Hauser LJ. 2010. Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinform. 11:119.
• Singh U, Wurtele ES. 2021. Orfipy: a fast and flexible tool for extracting ORFs. Bioinformatics 37:3019–3020.

8. Historic of versions

v1.1.0

• Receptor detection:

  • A filter of >65% coverage of the query HMM has been introduced to minimize false positives
  • The Com_TPR (PF18710) HMM has been added to the library of HMMs for TPRs

• RAM Usage:

  • the --chunk_size option has been introduced, which enables to divide the target dataset into chunks of N genomes in an effort to preserve RAM usage.

• Search options:

  • the --rbs_bins and start_codons options have been introduced, which enable to explictily specify the RBS motifs and start codons to consider for the detection of small ORFs encoded in the vicinity of candidate receptors

• Strictness levels:

  • QSSs with an unannotated propeptide preceded by a RBS_bin <= 13 are no longer included in the 'strict' output

v1.0.0

• Receptor detection:

  • The library of HMMs for TPRs has been updated using the HMMs present in interproscan-5.56-89.0
  • 20 additional HMMs of TPRs from Pfam have been included: TPR_1 (PF00515.30), TPR_2 (PF07719.19), TPR_3 (PF07720.14), TPR_4 (PF07721.16), TPR_5 (PF12688.9), TPR_6 (PF13174.8), TPR_7 (PF13176.8), TPR_9 (PF13371.8), TPR_10 (PF13374.8), TPR_11 (PF13414.8), TPR_14 (PF13428.8), TPR_15 (PF13429.8), TPR_16 (PF13432.8), TPR_17 (PF13431.8), TPR_18 (PF13512.8), TPR_19 (PF14559.8), TPR_20 (PF14561.8), TPR_21 (PF09976.11), TPR_22 (PF18833.3) and TPR_MalT (PF17874.3)
  • The tprpred software has been integrated and can be called in complement of hmmsearch to increase the sensitivity of the tool
  • An HMM of the AimR family has been built and is now used to identify these receptors with more sensitivity

• Propeptide detection:

  • The main improvement of v1.0.0 lies in a new algorithm to detect small peptides encoded in the vicinity of receptors, as such small peptides are typically absent from annotation files. We called this method SPRAT for Small Peptides with RBS Annotation Tool. This method identifies peptides preceded by a Shine-Dalgarno RBS in the flanking regions of each receptor, using the 27 hierarchical regular expressions introduced by Prodigal to detect SD-RBS motifs. This is justified by the fact that 90% of the canonical genes encoded by Firmicutes have an SD-RBS upstream. Optionally, the user can submit a list of possible start codons to consider for the detection of putative small pepite-coding ORFs (by default, only ATG is considered).
  • The PrediSi software has been integrated an can be called in complement of SignalP to increase the sensitivity of the tool
  • An HMM of SHP propeptides has been built and is now used to identify SHP propeptides since SHPs are not exported via the SEC-translocon and are therefore not returned by SignalP or PrediSi

• Iterative Search:

  • When the target database is large, it may be relevant to use the detected systems as baits to fish homologous systems that did not pass the conservative thresholds of RRNPP_detector. If this option increases the sensitivity of the tool, it also increases the risk of false positives

v0.0.1

Initial push of RRNPP_detector