Quick overview

Goal

Creation of a Genome-scale metabolic models (GEMs) of the outer retina (i.e. the RPE and PR layers) that can be used to apply constraint-based reconstruction and analysis (COBRA) methods to investigate hypotheses about eye metabolism.

Step 1: Creation of Cell-Specific GEMs

make_cs_models.ipynb Cell-specific models are constructed utilizing CORDA2 for Python within the corda5 environment.

Step 2: Creation of Retina Specific Metabolic Models

make_RPE-PR_models.ipynb This involves coupling RPE and rod/cone cell-specific models produced in Step 1 to generate retina-specific metabolic models.

Step 3: Analysis

analysis.ipynb The third step encompasses the analysis of the models created in the previous steps.

what you find in the 'models' folder

•    template model   contains template model (Human1)
•    cs_mods   contains cell-specific models generated in Step 1
•    models_outer_retina   contains retina-specific models generated in Step 2 and models used to generate them

More information

Environments

Before proceeding with the installation, make sure you have an environment management system (like Conda or Pipenv) installed on your system. You can download Conda here. These instructions are adaptable for various environment management systems.

Corda environment to create cell-specific GEMs using CORDA2

Follow these instructions in the Anaconda prompt to install the required environment to create of Cell-Specific GEMs using CORDA2:

  conda create --name corda
  
  pip install corda
  
  pip install openpyxl

RPE-PR environment to create of retina specific metabolic models and perform analysis

Follow these steps to install the required environment to couple RPE and PR cells and do the analysis:

  conda create --name RPE-PR

  pip install cobra

Creation of cell-specific GEMs

We created cell-specific models using an implementation of Cost Optimization Reaction Dependency Assessment (CORDA2; Schultz et al., 2017; Schultz & Qutub, 2016) for Python implemented by Christian Diener (https://github.com/resendislab/corda).

Template model

We used Human1 (Robinson et al., 2020) as template model (28-06-2023; https://github.com/SysBioChalmers/Human-GEM).

Single-cell expression data:

• RPE: Voigt et al., 2019 (Supplementary data, File pnas.1914143116.sd01.xlsx, column mean_expr_7)
• PR rod: Lukowski et al., 2019 (Supplementary data, Dataset EV1, File EMBJ-38-e100811-s003.xlsx, Average Rod PR C0, C1, C2, C3, C4, C7)
• PR cone: Lukowski et al., 2019 (Supplementary data, Dataset EV1, File EMBJ-38-e100811-s003.xlsx, C10 Cone PR)

1. Discretize expression into the following expression confidence levels:
   •   -1   for not present (< .00001)
   •   0   for unknown confidence (missing value)
   •   1   for low confidence (.00001 > Q25)
   •   2   for medium confidence (Q25 > Q75)
   •   3   for high confidence (> Q75)

            Note: for Lukowski PR rod and cone data most values fell into expression confidence levels -1, 2, or 3.

2. Mapping of expression confidence levels to reaction confidence levels

   • Using the reaction_confidence function from the CORDA package; "and" is evaluated by the minimum confidence and "or" by the maximum confidence.

3. Reconstruction of cell-specific models

   • objective function: biomass equation (so CORDA removes reactions in pathways that are not contributing to the production of biomass)
   • exchange bounds unchanged from downloaded Human1 model

Creation of the RPE-PR Models in COBRApy

1. Load PR and RPE Models.

2. Add Back Lost Exchange Reactions and ATP Hydrolysis Reaction.

3. Give Metabolite, Reaction, and Compartment IDs Unique Suffixes

   • _RPE or _PR depending on the cell-type.

4. Create RPE-PR Interface for the PR Model

   • Add RPE-PR interface reactions by duplicating reactions involving the extracellular space ([e_PR]) and assigning new suffixes:
     • Metabolite ID suffix: e_RPE_PR
     • Reaction ID suffix: _PR-RPE
     • Compartment: [e_PR] -> [e_RPE_PR]

5. Create RPE-PR Interface for the RPE Model

   • Add RPE-PR interface reactions by duplicating reactions involving the extracellular space ([e_RPE]) and assigning new suffixes:
     • Metabolite ID suffix: e_RPE_PR
     • Reaction ID suffix: _RPE-PR
     • Compartment: [e_RPE] -> [e_RPE_PR]

6. Fuse RPE and PR Models

   # fuse models
   mod_RPE_PR = mod_RPE.copy()
   mod_PR_copy = mod_PR.copy()
   mod_RPE_PR.id = mod_RPE.id + '_' + mod_PR.id
   mod_RPE_PR.name = mod_RPE.id + '_' + mod_PR.id
   mod_RPE_PR.add_reactions(mod_PR_copy.reactions)

7. Delete duplicated reactions

   • Delete one duplicate of duplicated interface reactions (that solely take place in the [e_RPE_PR] compartment), and change reaction IDs of remaining reaction to _eRPE_PR

8. Fix compartment dictionary

   • Make sure all compartment dictionary keys are associated to values (otherwise SBML model is not valid).

9. Make sure added reactions are added to their 'subsystem' groups

    # add all reactions to their subsystem group
    g_dict = {g.name:g.id for g in model.groups}
    for r in model.reactions:
        model.groups.get_by_any(g_dict[r.subsystem])[0].add_members([r])

10. Save model in SBML format

A schematic overview

Some of the results

## References Diener, C. (2023, September 1). CORDA for Python. https://github.com/resendislab/corda

Lukowski, S. W., Lo, C. Y., Sharov, A. A., Nguyen, Q., Fang, L., Hung, S. S., Zhu, L., Zhang, T., Grünert, U., Nguyen, T., Senabouth, A., Jabbari, J. S., Welby, E., Sowden, J. C., Waugh, H. S., Mackey, A., Pollock, G., Lamb, T. D., Wang, P., … Wong, R. C. (2019). A single-cell transcriptome atlas of the adult human retina. The EMBO Journal, 38(18). https://doi.org/10.15252/EMBJ.2018100811

Robinson, J. L., Kocabaş, P., Wang, H., Cholley, P. E., Cook, D., Nilsson, A., Anton, M., Ferreira, R., Domenzain, I., Billa, V., Limeta, A., Hedin, A., Gustafsson, J., Kerkhoven, E. J., Svensson, L. T., Palsson, B. O., Mardinoglu, A., Hansson, L., Uhlén, M., & Nielsen, J. (2020). An atlas of human metabolism. Science Signaling, 13(624). https://doi.org/10.1126/SCISIGNAL.AAZ1482

Schultz, A., Mehta, S., Hu, C. W., Hoff, F. W., Horton, T. M., Kornblau, S. M., & Qutub, A. A. (2017). Identifying cancer specific metabolic signatures using constraint-based models. Pacific Symposium on Biocomputing, 0, 485–496. https://doi.org/10.1142/9789813207813_0045

Schultz, A., & Qutub, A. A. (2016). Reconstruction of Tissue-Specific Metabolic Networks Using CORDA. PLOS Computational Biology, 12(3), e1004808. https://doi.org/10.1371/JOURNAL.PCBI.1004808

Voigt, A. P., Mulfaul, K., Mullin, N. K., Flamme-Wiese, M. J., Giacalone, J. C., Stone, E. M., Tucker, B. A., Scheetz, T. E., & Mullins, R. F. (2019). Single-cell transcriptomics of the human retinal pigment epithelium and choroid in health and macular degeneration. Proceedings of the National Academy of Sciences of the United States of America, 116(48), 24100–24107. https://doi.org/10.1073/PNAS.1914143116

Contact

Get in touch with me if you have any questions or suggestions!

Stella: s.prins@ucl.ac.uk

Institute of Ophthalmology, University College London, London, UK

Advanced Research Computing, University College London, London, UK

This work was funded by Moorfields Eye Charity.

Consider donating.