Phase 3: condition-specific proteome-constrained ecModels

.github/workflows/ci.yml

@@ -22,4 +22,4 @@ jobs:
           python -m pip install --upgrade pip
           pip install -e ".[dev]"
       - name: Test
-        run: pytest -v --durations=10
+        run: pytest -v --durations=10 -m "not integration"

README.md

@@ -59,6 +59,35 @@ model = load_model(build_adapter(conditions["CN4"]))
 `build_adapter` reads `model_adapter.toml` and layers the condition's measured
 protein content and dilution rate on top.
 
+## Building the condition models
+
+```bash
+python -m overflow.build                 # every condition
+python -m overflow.build CN4 --solver glpk
+```
+
+Each condition gets `models/ecModel_P_<cond>.yml` and, in `results/`, the
+enzymes whose measured abundance the model could not run on, the enzymes whose
+caps had to be released, and the full flux distribution.
+
+Options change what the models are asked to account for:
+
+- `--fit-rates` holds CO2, oxygen and the byproducts within a tolerance of their
+  measurements (`--rate-tolerance`, default 5%) and raises the measured abundances
+  by the least that makes that feasible. Without it those rates are free, and the
+  model disposes of surplus carbon through whichever exit is cheapest in protein
+  rather than respiring it.
+- `--uptake-flex` caps glucose uptake at that multiple of the measured rate
+  (default 1.05). CN4 needs 1.08: at its dilution rate the model needs 6.5% more
+  glucose than was measured, even with protein unlimited.
+- `--objective` chooses what the finished model optimises: the smallest protein
+  pool (`protein`, the default), or the smallest total flux at the dilution rate
+  over every reaction (`flux`) or over the metabolic reactions only
+  (`metabolic-flux`). The flux objectives need `--fit-rates`.
+- `--scale-protein` rescales biomass to the measured protein content. Off by
+  default: the model has no carbon to spare at the measured glucose uptake, and
+  CN4 then falls short of its dilution rate.
+
 ## Tests
 
 ```bash

data/README.md

@@ -0,0 +1,22 @@
+# Data
+
+| File | Contents |
+|---|---|
+| `abs_proteomics.txt` | Absolute protein abundances [mmol/gDW]. 2929 rows, one header; three biological replicates per condition and four for hGR. Missing measurements are `NA`. |
+| `fermentationData.txt` | Per condition: total protein content [g/gDW], dilution rate [1/h] and measured exchange rates [mmol/gDW/h]. An undetected byproduct is `NA`. |
+| `ribosome.txt` | Ribosomal subunits with gene names, mass and sequence, from UniProt. |
+| `selectedAnnotation.txt` | Assignment of proteins to the systems used when summarising enzyme usage. |
+| `uniprot.tsv` | Yeast proteome with molecular masses, used to convert abundances from mmol/gDW to mg/gDW. |
+
+## abs_proteomics.txt
+
+`P61830` appears on two rows, once per gene of the duplicated histone
+(`YBR010W`, `YNL031C`), with identical abundances. Readers collapse it to one
+entry and fail if the duplicated rows ever disagree.
+
+## uniprot.tsv
+
+From [SysBioChalmers/GECKO](https://github.com/SysBioChalmers/GECKO),
+`tutorials/full_ecModel/data/uniprot.tsv`, branch `develop4`, commit
+`cbc4ca3311df8ca44c28792f40008b9cbef692bd`. UniProt proteome UP000002311,
+reviewed entries. Covers every protein in `abs_proteomics.txt`.

models/PROVENANCE.md

@@ -6,15 +6,16 @@ The full enzyme-constrained yeast model, in GECKO 4 / RAVEN 3 YAML format.
 
 | | |
 |---|---|
-| Source | [SysBioChalmers/GECKO](https://github.com/SysBioChalmers/GECKO), `tutorials/full_ecModel/models/ecYeastGEM.yml` |
-| Branch | `develop4` |
-| Commit | `cbc4ca3311df8ca44c28792f40008b9cbef692bd` |
-| Blob | `891ad7ba769cdc7d0fd4b605b0fd625c8cdd6867` |
-| Built with | GECKO 4.0.0b1, RAVEN 3.0.0b1 |
+| Source | [SysBioChalmers/ecModels](https://github.com/SysBioChalmers/ecModels), `ecYeastGEM/models/ecYeastGEM.yml` |
+| Branch | `gecko4` |
+| Commit | `880288747ad98220764ce07f8b7a396048e0537d` |
+| Blob | `d684eb01ddaf49da9ac9f5891aae372e16b34542` |
+| Base model | yeast-GEM 9.1.1 |
+| Built with | geckopy 4.0.0b1 |
 
-1144 enzymes over 4834 enzyme-constrained reactions. kcat sources: 3280 BRENDA,
-1076 DLKcat, 251 standard, 217 custom, 6 sensitivity-tuned, 3 isozyme-derived,
-1 set explicitly. Subunit stoichiometries come from Complex Portal.
+1144 enzymes over 4850 enzyme-constrained reactions. The kcats were tuned
+against measured growth rates and exchange fluxes with CMA-ES, see the
+`ecYeastGEM` README in the source repository.
 
 The model ships with `r_4046` (non-growth associated maintenance) pinned at
 0.7 mmol ATP/gDW/h and `prot_pool_exchange` bounded at 125 mg/gDW, which is
@@ -23,14 +24,17 @@ per condition when building the proteome-constrained models.
 
 ## yeast-GEM.yml
 
-The conventional GEM the ecModel was built from, from the same commit. Used
-as `conv_gem` by the geckopy adapter and as the starting model for random
+The conventional GEM the ecModel was built from, yeast-GEM release `v9.1.1`
+(`model/yeast-GEM.yml`, sha256
+`2ddfaaa73ead41ab243127d1b34fa56db2c86052f6eabcd12bb4eea93947dc17`). Used as
+`conv_gem` by the geckopy adapter and as the starting model for random
 sampling.
 
 ## Verifying
 
 ```bash
-curl -sL https://raw.githubusercontent.com/SysBioChalmers/GECKO/cbc4ca3311df8ca44c28792f40008b9cbef692bd/tutorials/full_ecModel/models/ecYeastGEM.yml \
+curl -sL https://raw.githubusercontent.com/SysBioChalmers/ecModels/880288747ad98220764ce07f8b7a396048e0537d/ecYeastGEM/models/ecYeastGEM.yml \
   | git hash-object --stdin
-# 891ad7ba769cdc7d0fd4b605b0fd625c8cdd6867
+# d684eb01ddaf49da9ac9f5891aae372e16b34542
+curl -sL https://raw.githubusercontent.com/SysBioChalmers/yeast-GEM/v9.1.1/model/yeast-GEM.yml | sha256sum
 ```