henhao
Sign in
BioinformaticsNature Biotechnology
Peer-reviewed journal article

The DOE Systems Biology Knowledgebase (KBase)

Adam P. ArkiniD, Robert W. CottinghamiD, Christopher S. HenryiD, Nomi L. HarrisiD, Rick L. StevensiD, Sergei MasloviD, Paramvir DehaliD, Doreen WareiD, Fernando PéreziD, Richard Shane CanoniD, Michael W. Sneddon, Matthew L Henderson, William J. RiehliD, Dan Murphy-Olson, Stephen Y. ChaniD, Roy T. KamimuraiD, Sunita KumariiD, Meghan M. DrakeiD, Thomas Scott BrettiniD, Elizabeth M. Glass, Dylan C. ChivianiD, Dan GunteriD, David J. WestoniD, Benjamin H. AlleniD, Jason K. Baumohl, , Dantong Yu

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

Abstract

The U.S. Department of Energy Systems Biology Knowledgebase (KBase) is an open-source software and data platform designed to meet the grand challenge of systems biology -- predicting and designing biological function from the biomolecular (small scale) to the ecological (large scale). KBase is available for anyone to use, and enables researchers to collaboratively generate, test, compare, and share hypotheses about biological functions; perform large-scale analyses on scalable computing infrastructure; and combine experimental evidence and conclusions that lead to accurate models of plant and microbial physiology and community dynamics. The KBase platform has (1) extensible analytical capabilities that currently include genome assembly, annotation, ontology assignment, comparative genomics, transcriptomics, and metabolic modeling; (2) a web-browser-based user interface that supports building, sharing, and publishing reproducible and well-annotated analyses with integrated data; (3) access to extensive computational resources; and (4) a software development kit allowing the community to add functionality to the system.

02 /Specification & Classification

Click any tag to browse matching papers
Citations
1,624
FWCI
160.54
Field percentile
top 1%
OpenAlex Topics

In-depth summary

✦ Multi-agent verified full-text synthesis

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (22 Dec 2016) · CC BY 4.0

Open-source KBase platform unites reference genomes and biochemistry, over 70 apps, Jupyter-based Narratives with provenance, and a Docker SDK for reproducible plant and microbial systems biology.

Turning heterogeneous, error-prone genomic, expression, metabolite and phenotype data into predictive models takes many tools and close collaboration, yet existing platforms are decentralized and each lacks modeling, genome annotation, or integrated reference data. KBase links one data model, full provenance and shareable Narratives so that user and reference data are analyzed with the same tools.

Key results

  1. ShownKBase combines curated reference data (public genomes, a biochemistry database and major ontologies) with user uploads and a catalog of interoperable apps that run from read assembly through metabolic and community modeling.

    Reference set: all public RefSeq and Phytozome genomes plus selected Ensembl plant genomes; 27,692 compounds, 34,705 reactions and 529 growth media; GO, SEED Subsystems, PFam and Interpro. Over 70 production apps plus dozens of beta apps. · KBase Data Model and Apps

  2. ShownPublic Narratives are cited by seven peer-reviewed studies, most of whose analysis was run in KBase. The studies cover plant and microbial metabolic reconstruction, microbiome community modeling, phenotype-driven comparative genomics and expression analysis.

    PlantSEED models for 10 reference plant genomes; core models for more than 8000 microbial genomes, compared phylogenetically; community models of gut, cyanobacterium-heterotroph and electrosynthetic consortia (three dominant species); TN-seq check of predicted gene essentiality; co-expression clusters in five genomes. · Science Performed Within the KBase Platform; Tables

  3. ShownUsage since the Narrative interface launched in February 2015 shows active use, although only a small share of user Narratives had been shared or made public.

    More than 1200 registered users, over 750 of whom had made at least one Narrative. Non-staff users built 2657 Narratives: 294 shared with another user and 37 public, holding 255,148 data objects (average of 96 objects and five apps per Narrative). · Science Performed Within the KBase Platform

  4. SuggestedIn a worked example, a wet-lab scientist and a modeling scientist collaborated across five Narratives using real Shewanella amazonensis SB2B reads and Biolog phenotype data. The result was a new SB2B genome-scale metabolic model and a refined S. oneidensis MR-1 model.

    Steps across Table 1 Narratives 1 to 5: assemble and annotate reads, compare with related Shewanella strains, propagate a published MR-1 model, fit models to Biolog growth data, and swap gap-filled reactions for biologically justified ones. No quantitative fit statistics are given in the text. · Support for Reproducible, Interdisciplinary…; Tables

  5. SuggestedA Docker-based SDK lets any developer build, test, register and deploy tools as versioned apps, while the platform handles provenance, job management and logging. The authors state that all current KBase tools were built with it.

    Containers run the same way on a developer's machine as in production, and they deploy once registered. Contributed code must carry a standard open-source license. One published study used the SDK to add a co-expression clustering app (Table 2, Narrative 7). Platform code is MIT-licensed. · KBase Software Development Kit; Science Performed Within the KBase Platform; Code Availability

Methods, data and reproducibility

Service-oriented platform: a Jupyter-based Narrative interface submits jobs to an Execution Engine, which routes Docker-containerized apps to an OpenStack cloud and DOE HPC. An object store with typed references tracks versioning, provenance, permissions and three-level sharing across user and reference data.

  • Reference genomes: all public RefSeq and Phytozome genomes plus selected Ensembl plant genomes, with original and KBase-updated annotations · multiple (plants and microbes)
  • Reference biochemistry: compounds, reactions and growth media formulations, used for metabolic modeling · n = 27,692 compounds; 34,705 reactions; 529 growth media
  • Raw reads from an existing Shewanella amazonensis SB2B genome, plus Biolog growth phenotype data from a prior study (Alice and Bob example) · Shewanella amazonensis SB2B
  • Published S. oneidensis MR-1 metabolic model and its growth phenotype data, used for model propagation and comparison · Shewanella oneidensis MR-1
  • Reference plant genomes reconstructed and compared with the PlantSEED pipeline (Table 2, Narrative 1) · plants (multiple species) · n = 10
  • Microbial genomes used for core metabolic model reconstruction and phylogenetic comparison (Table 2, Narrative 2) · microbes (multiple species) · n = more than 8000
  • Community modeling cases: a gut pair, a cyanobacterium plus heterotroph with metatranscriptomes, and an electrosynthetic microbiome (metagenome and metatranscriptome) · B. thetaiotaomicron, F. prausnitzii, T. elongatus, M. ruber, others · n = three dominant species in electrosynthetic microbiome
  • Expression data used to find and compare co-expressed gene clusters (Table 2, Narrative 7) · E. coli, S. oneidensis, P. aeruginosa, T. thermophilus, S. aureus · n = five genomes
Methods & tools
Narrative Interface (built on Jupyter Notebook), KBase Software Development Kit (SDK), Docker, OpenStack, PlantSEED, Flux balance analysis, Gap-filling (including growth phenotype gap-filling), Biolog growth phenotype arrays, RefSeq, Phytozome, Ensembl, GO, SEED Subsystems, PFam, Interpro
Code
https://github.com/kbasehttps://narrative.kbase.us/#appcatalog
Data & accessions
http://www.kbase.us/narrative-libraryhttp://www.kbase.us/kbase-paperhttp://www.kbase.us/community-modeling/
Software & environment
Each app and its dependencies are packaged as a Docker image, versioned and served by the App Catalog. Contributed software must use a standard open-source license, and KBase code is under the MIT License. No package versions are listed.
Assay conditions & compute
OpenStack cloud at Argonne and Lawrence Berkeley National Laboratories, plus NERSC and ALCF supercomputers; HPC integration was only beginning. No runtimes or memory footprints are reported.
Reproducibility gaps
Viewing the Table 1 and 2 Narratives requires signing up for a KBase account. Specific Narrative links appear only in the tables or online, not in the text, and no dataset accessions are printed.
Limitations (6)
  • External developers are not yet able to add new data types to the KBase data model; the authors call this a near-term goal. · stated by the authors
  • Faceted search covered only reference data; extending it to users' own and shared data while protecting privacy was still future work. · stated by the authors
  • HPC integration was only beginning. The authors list bulk data input, large-scale cloud and HPC execution, metagenomics, eukaryotic genomics and comparative genomics as areas still needing expansion. · stated by the authors
  • Programmatic access through Narrative code cells was still a prototype, with its API under construction. Formal Narrative publication and project-level sharing were planned but not yet implemented. · stated by the authors
  • The claim that KBase alone combines community modeling with expression comparison, and the claim that users learn multi-step analyses quickly, rest on author statements and unquantified user feedback, not a feature benchmark or a user study. · noted in this summary
  • Only 294 of 2657 non-staff Narratives were shared and 37 made public, so the evidence for collaborative reuse and reproducibility comes mostly from a few curated example Narratives, not broad usage data. · noted in this summary

Who should read:Read this if you build or use web platforms for reproducible microbial and plant genomics, metabolic or community modeling, and want KBase's design, data model, provenance system and SDK.

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
THISVega '20Launer-Wach '22Shaffer '23Ma '26Bruskiewich '16Martinez-Ve '17
60% semantic matchSystems Biology
Carlos Vega, Valentin Grouès, Marek Ostaszewski+2 more·bioRxiv·Oct 1, 2020· 1 citations