henhaobio
Sign in

bioRxivBioinformaticsPreprint

PLaBAse: A comprehensive web resource for analyzing the plant growth-promoting potential of plant-associated bacteria

Sascha PatziD, Anupam GautamiD, M. BeckeriD, Silke RuppeliD, Pablo Rodríguez‐PalenzuelaiD, DH. Huson

PDFDOI
In-depth full-text digest available · Jump to digest ↓

Abstract

Plant-beneficial microorganisms are gaining importance for sustainable plant production and phytosanitary practices. Yet there is a lack of computational approaches targeting bacterial traits associated with plant growth-promotion (PGP), which hinders the in-silico identification, comparison, and selection of phytostimulatory bacterial strains. To address this problem, we have developed the new web resource PLaBAse (v1.01, http://plabase.informatik.uni-tuebingen.de/pb/plabase.php), which provides a number of services, including (i) a database for screening 5,565 plant-associated bacteria (PLaBA-db), (ii) a tool for predicting plant growth-promoting traits (PGPTs) of single bacterial genomes (PGPT-Pred), and (iii) a tool for the prediction of bacterial plant-association by marker gene identification (PIFAR-Pred). The latter was developed by Martinez-Garcia et al. and is now hosted at University of Tuebingen. The PGPT-Pred tool is based on our new PGPT ontology, a literature- and OMICs-curated, comprehensive, and hierarchical collection of [~]6,900 PGPTs that are associated with 6,965,955 protein sequences. To study the distribution of the PGPTs across different environments, we applied it to 70,540 bacterial strains associated with (i) seven different environments (including plants), (iii) five different plant spheres (organs), and (iii) two bacteria-induced plant phenotypes. This analysis revealed that plant-symbiotic bacteria generally have a larger genome size and a higher count of PGPT-annotated protein encoding genes. Obviously, not all reported PGPTs are restricted to -or only enriched in-plant-associated and plant symbiotic bacteria. Some also occur in human- and animal-associated bacteria, perhaps due to the transmission of PGP bacteria (PGPBs) between environments, or because some functions are involved in adaption processes to various environments. Here we provide an easy-to-use approach for screening of PGPTs in bacterial genomes across various phyla and isolation sites, using PLaBA-db, and for standardized annotation, using PGPT-Pred. We believe that this resource will improve our understanding about the entire PGP processes and facilitate the prediction of PGPB as bio-inoculants and for biosafety strategies, so as to help to establish sustainable and targeted bacteria-incorporated plant production systems in the future.

In-depth summary

Claude Opus 5.5 · xhigh effort · 17 Oct 2026 · from the full text of version 1 (15 Dec 2021)

PLaBAse annotates bacterial genomes against a curated ontology of ∼6,900 plant growth-promoting traits; applied to 70,540 genomes, plant symbionts and rhizobacteria had the largest genomes and most traits.

Few computational approaches target the bacterial traits behind plant growth promotion, which makes it hard to identify, compare and select candidate bio-inoculant strains in silico. General annotation resources (KEGG, GO, eggNOG) and earlier plant-microbe work such as PIFAR's ∼170-trait predictor do not treat these traits as one entity, whereas PLaBAse gathers plant growth-promoting traits (PGPTs) into a hierarchical ontology linked to 6,965,955 reference proteins.

Key results

  1. ShownPLaBAse bundles a database of 5,565 annotated plant-associated strains, a PGPT annotation tool (PGPT-Pred) and a plant-association predictor (PIFAR-Pred), built on an ontology of 6,912 traits mapped to 12,593 leaf nodes across seven levels.

    PGPT-Pred aligns submitted proteins to 6,965,955 PGPT reference sequences (blastp, e-value 1e-5, plus a hmmer Pfam check) or maps IMG KEGG annotations (6,222 PGPTs carry a K number). Ontology: 3 classes at level 1, 8 at level 2, 47 at level 3, 280 at level 4, 1,966 at level 5. PIFAR-Pred is the Martínez-García et al. tool; result pages stay online for 15 days. · Web Server setup for PLaBAse; The PLaBAse-Web resource and services; The PGPT Ontology System; Environments: Higher PGPT counts mostly…

  2. ShownPlant- and soil-associated bacteria carry significantly more PGPTs per genome than isolates from other environments, and plant symbionts more than plant pathogens, while PGPTs make up roughly one third of genes in every environment.

    Level 0 averages: plant-associated (PA) ∼1,749 and soil-associated (SA) ∼1,559 PGPTs per genome, above all other environments; rhizospheric ∼1,943 vs anthospheric ∼1,314; plant symbionts ∼2,139 vs pathogens ∼1,725 (∼1,300 when human and animal pathogens are added). Some human-, food- and industry-associated strains still reach PA-like counts. · Environments: Higher PGPT counts mostly…; Plant spheres: Highest PGPT counts are strongly…; Bacteria-induced plant phenotypes: On average…

  3. ShownGenome size follows the same pattern: plant- and soil-associated bacteria, above all rhizosphere strains and plant symbionts, have larger genomes and more genes than human-, animal-, aquatic-, food- or industry-associated bacteria.

    PA and SA average 5,312 – 5,344 genes (5,576,974 – 5,665,934 nt) vs 3,869 (human), 3,201 (animal), 4,113 (aquatic), 3,482 (food) and 4,492 (industrial); rhizospheric strains 5,913 genes; plant symbionts 6,443 genes (6,644,481 bp) vs pathogens 5,090 (5,447,664 bp). One-way ANOVA with TukeyHSD, p-value < 0.05; complete genomes alone gave similar or stronger patterns. · Genomic Data Acquisition and Classification; Plant- and soil associated bacteria have higher…; Across different plant spheres the higher genome…; Overall highest genome sizes are found in…

  4. ShownEnrichment tests pick out PGPTs specific to plant-associated and symbiotic bacteria: many plant-enriched traits are also enriched in soil bacteria, and symbiont-enriched traits are dominated by colonization and competitive-exclusion functions.

    Scoary with Fisher's exact tests and Benjamini-Hochberg correction. Level 6: 2,188 of 4,855 significantly altered PGPTs enriched in PA (odds ratio ≥1.5), 1,357 of them also in SA, 866 in PA only; up to 74 traits absent from PA. Symbionts: 1,821 enriched PGPTs not shared with the 1,186 pathogen-enriched ones, 393 of them not enriched in PA; colonization 30%, competitive exclusion 21%. · Bacterial Genome-Wide PGPT Enrichment Study; Environmental specific clustering of PGPTs…; Plant-symbiotic and -pathogenic bacteria reveal…

  5. SuggestedPGPT class counts largely, but not completely, separate plant symbionts from plant pathogens: individual symbiotic and pathogenic strains fall into both clusters.

    77% of the variance in PGPT count between symbionts and pathogens is attributed to higher assignments across all level 3 classes and 9% to mostly putative functions (supplementary figures; the analysis behind these percentages is not described in this text). Symbionts were compared with pathogens as 855 vs 890 strains. · Genomic Data Acquisition and Classification; PGPTs facilitate differentiation between Plant…

Methods, data and reproducibility

PGPTs curated from literature and databases were organised into a seven-level multi-labelled tree backed by 6,965,955 non-redundant reference proteins (uclust) and Pfam domains; genomes are annotated by blastp (e-value 1e-5) plus hmmer or by mapping IMG KEGG annotations, and IMG genomes were compared by isolation site, plant sphere and phenotype with ANOVA/TukeyHSD and Scoary enrichment.

  • Bacterial genomes and metadata from the IMG database, downloaded November 2020 · bacteria (many phyla) · n = 71,931 downloaded; 70,540 with at least 450 genes; 38,735 assigned to groups
  • Isolation-site groups: plant, human, animal, soil, aquatic, food and industrial-associated strains · bacteria · n = 5,565 PA, 17,046 HA, 5,065 AA, 5,464 SA, 5,656 QA, 1,190 FA, 1,166 IA
  • Plant-sphere subsets of plant-associated strains (rhizosphere, phyllosphere, anthosphere, carposphere, spermosphere) · bacteria · n = 2,261, 708, 31, 55 and 134 strains
  • Strains with a known bacteria-induced plant phenotype · bacteria · n = 855 plant symbiotic, 890 phytopathogenic
  • PGPT reference proteins gathered from KEGG, GO, eggNOG, UniProt and Meta-/Bio-Cyc, plus MIBiG compounds · bacteria · n = 6,965,955 protein sequences
Methods & tools
PLaBAse v1.01, PLaBA-db, PGPT-Pred v1.1.0, PIFAR-Pred v1.2.0, blastp (v2.10.1+), HMMER 3.3, IMG-KEGG-annotation_MAPPER, uclust v1.2.22q, Scoary v.1.6.16, Pfam-A, KronaTools v2.8 and Plotly, MySQL
Reagents & environment
Web server back end: PHP v7.2.24, Perl v5.26.1, Python3 v3.7.4, R v4.0.2, MySQL v5.7.34; blastp v2.10.1+, HMMER v3.3, uclust v1.2.22q, Scoary v.1.6.16, Plotly v1.58.5, KronaTools v2.8.
Reproducibility gaps
The PLaBAse web address and the Pfam ftp link were lost in extraction; no source-code repository or ontology download is named in this text, the Scoary command is not shown, and result pages expire after 15 days.
Limitations (6)
  • Plant symbionts and growth-promoting bacteria cannot yet be told apart by phenotype or genotype, so they are pooled and no prediction tool for growth-promoting bacteria is offered. · stated by the authors
  • Archaeal traits are excluded and not all plant-beneficial pathways are known, so the ontology is incomplete and more in vivo work is needed. · stated by the authors
  • Colonization and competitive-exclusion traits may also serve other hosts and environments, and bacteria switching lifestyles can blur symbiont-pathogen separation by trait counts. · stated by the authors
  • Rhizosphere-specific enrichment may be biased by the larger number of rhizospheric strains. · stated by the authors
  • Trait calls rest on sequence homology or KEGG mapping plus IMG isolation-site labels; no experimental validation or accuracy test of PGPT-Pred against strains with known growth-promoting phenotypes is reported. · noted in this summary
  • Group sizes are very unbalanced (17,046 human- vs 1,166 industrial-associated strains; 31 anthosphere isolates), which can sway averages and enrichment tests beyond the rhizosphere case the authors flag. · noted in this summary

Read this if you screen bacterial isolates or genomes for plant growth-promoting potential, or study how plant-beneficial gene repertoires differ between environments, plant organs, symbionts and pathogens.

Citations
152

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISPatz '24Poncheewin '22Pille '24Petrillo '21Belmonte '20Cotta '20
78% semantic matchBioinformatics
Sascha Patz, Mario Rauh, Anupam Gautam+1 more·bioRxiv·Feb 17, 2024· 25 citations