Update MLR estimates

Run via command line script run-mlr-model.py rather than Jupyter notebook

.gitignore

@@ -16,6 +16,9 @@ metadata.tsv.gz
 predictors/nextclade_sars-cov-2_21L.json
 predictors/nextclade_sars-cov-2_21L.json.gz
 
+mlr-fitness/estimates/models/
+mlr-fitness/estimates/pango_results.json
+
 # Compiled LaTeX output
 ncov-escape.pdf
 

mlr-fitness/estimates/README.md

@@ -1,9 +1,20 @@
 # MLR fitness estimates using evofr
 
-1. Clone [evofr](https://github.com/blab/evofr) repo elsewhere in the file system and install via
+1. Install latest version of [evofr](https://github.com/blab/evofr) via pip:
 ```
-poetry build
-pip install <path-to-wheel>
+pip install evofr
 ```
 
-2. Run the notebook `evofr_mlr.ipynb` via `jupyter notebook`. This will export a file called `growth_advantages.tsv` that contains the median growth advantage and 80% credible interval per Pango lineage.
+2. Run script `run-mlr-model.py` via:
+```
+python -u run-mlr-model.py --config mlr-config.yaml --export-path . --data-name "pango"
+```
+
+This produces the file `pango_results.json`.
+
+3. This file includes unnecessary information of full frequencies trajectories. The script `prune-mlr-results.py` walks through this JSON and extracts just the useful `site` of `ga`:
+```
+python prune-mlr-results.py --results pango_results.json > growth_advantages.tsv
+```
+
+And exports the TSV file called `growth_advantages.tsv` that contains the median growth advantage and 80% credible interval per Pango lineage.