Feb 11, 2021

Build Metabolic Model and Other FBA Apps Updated

We are excited to announce a major update to our popular Build Metabolic Model App and related tools. With the recent updates to the RAST annotation apps and the release of the new version of the ModelSEED Biochemistry Database (data on GitHub), we have updated our model templates to reflect the changes in those data sources at the core of the Model Reconstruction pipeline. The reconstruction methodology was improved to address issues with unconstrained production of energy in some organisms. Note that this app is for microbes only. For fungi and plants, model reconstruction remains available via the Build Fungal Model and Reconstruct Plant Metabolism Apps respectively.  Additionally, we have taken this opportunity to release smaller updates and bug fixes to other Apps in our suite of Metabolic Modeling tools.

What does this release mean for my current models and analysis?

All current models built prior to this release won’t suffer any changes. Only newly built models will reflect the changes in this release. If you are finalizing a study and want to keep using the previous version data and reconstruction pipeline, the previous version templates and “classic” pipeline for reconstruction are still available. Please check the Build Metabolic Model App documentation for full release notes to learn about the new features and how to access previous version templates and the classic reconstruction pipeline. 

If you annotated a genome with any of the RAST v1.8.1 apps (released on Oct. 13, 2020.) and started a model reconstruction analysis, we recommend you rebuild the model with the template version in this release. The new template is optimized for the updated RAST annotations. 

Release Notes:

Documentation for metabolic modeling related apps has been updated to provide additional information on app parameters and outputs. 

For details in changes, new features and improvements of individual apps see the detailed list below.

Build Metabolic Model

With the recent updates to the RAST annotation apps and a new release of the ModelSEED Biochemistry Database (insert link) we have updated the model reconstruction templates. The reconstruction pipeline relies on the mapping between functional roles annotated by RAST to the biochemical reactions in the ModelSEED DB (insert link). With the update to both data sources, the templates were updated to reflect those changes.  ATP production was improved in our model reconstruction procedure by constructing core models, testing for proper ATP production from this core, then ensuring that ATP production does not incorrectly explode when expanding the core model to a genome-scale model. We similarly improved our gapfilling approach to ensure that gapfilling does not cause a model to start over-producing ATP. While other approaches aim to correct ATP overproduction in models, these new procedures in the ModelSEED pipeline aim to ensure that ATP overproduction does not happen in the first place. New parameters have also been added, more details below:

  • Changes: The templates for model reconstruction have been updated. 
    • The ability to reconstruct a model for plants has been moved, and it is now available in the Reconstruct Plant Metabolism App.
    • The gram-negative and gram-positive templates have been updated. The “Automatic selection” will use the updated templates. 
    • The old templates (legacy) can be manually selected in the parameter dropdown. It is important to note that models built using the older templates will lead to different results if your genome(s) was annotated with the latest version of RAST. We do not recommend this practice. If you are trying to use older templates for reproducibility purposes please use a genome annotated with the older version of RAST.
    • All templates are available on Github
  • New parameter: Classic mode? (OFF by default)
    • Run the classic model reconstruction pipeline rather than the new pipeline that prevents the overproduction of ATP.
  • New parameter: Use internal annotations (ON by default)
    • Use the annotated functional roles in the genome as the primary annotation source (must be RAST annotations at this time).
      • This parameter was introduced to support the use of other annotations sources in future releases. 
  • New parameter: Merge all selected annotations (OFF by default)
    • Merge all selected annotations mapping all associated reactions to genes even if annotation sources disagree on function.
      • This parameter was introduced to support the use of other annotations sources in future releases. 
  • New parameter: Prioritized annotation sources
    • Use “Merge all selected annotations” is turned ON  you can select the available annotation sources in this genome to use in model reconstruction in prioritized order.

    Run Flux Balance Analysis             

    • Improvements: 
      • Gene Knockouts functionality now allows search of genes associated with the model removing the need to manually type gene names.

    Build Multiple Metabolic Models

    • Improvements:
      • In addition to individual genomes, genome sets can now be used as an input for model reconstruction

    Edit Metabolic Model

    • Improvements:
      • Biochemistry DB is now searchable when adding new additions to the model.

    View Flux Network

    • App has been deprecated.

         

        About the Authors

        José Faria
        José Faria

        José P. Faria is a Computational Biologist at Argonne’s recently formed Data Science and Learning division. He started his education as a Biologist and went on to pursue a Ph.D. in Bioengineering under the scope of the MIT Portugal Progam. He has been a member of the Henry Lab at Argonne since its inception in 2009, starting as a Visiting Fullbright Scholar to perform research for his Master’s thesis. In his young career, his research has focused on genome-scale metabolic modeling reconstruction and analysis for prokaryotes as a member of the ModelSEED team. He brought his expertise in metabolic modeling to KBase and actively engages in the development of new scientific tools for the community. In KBase he can also be found on the road as a member of the outreach team. He has lectured over 20 workshops in the last 4 years in 5 different countries. His research interests intersect the fields of Computational Biology, Bioinformatics, Metabolic Engineering and Systems Biology.

        He can be found on LinkedIn, Twitter and Google Scholar

        Ben Allen
        Ben Allen

        Ben Allen coordinates outreach and user development activities to build the KBase user community while engaging in scientific collaborations to advance the use of the platform. His background in biochemistry and science education helps him develop protocols and training materials that provide depth while being accessible to a wide audience. Research interests include systems biology, microbial ecology, bioremediation studies, and biology education.

        System maintenance

        On Wednesday, October 27, we will be performing system maintenance from 11am-1pm PDT (1-3pm CDT).  narrative.kbase.us and appdev.kbase.us may be unavailable or unstable, and compute jobs may fail, during this time.