Skip to content
This repository was archived by the owner on May 8, 2020. It is now read-only.

README.md

Latest commit

 

History

History
83 lines (55 loc) · 3.04 KB

README.md

File metadata and controls

83 lines (55 loc) · 3.04 KB

Epidemic Forecasting library and tools

Note: Track issues for this repo in https://github.com/epidemics/covid/issues

Libraries and utilities for data handling, estimation and modelling of epidemics. Part of https://github.com/epidemics/ project.

  • Needs Python 3.7+, git, poetry.
  • Install requirements with poetry install.

Quickstart

Install / preparation

  • First, install Poetry
  • Then, install the lib and its dependencies (e.g. in virtualenv, or let poetry create it)
git clone https://github.com/epidemics/epifor
cd epifor

poetry install
  • If you want to upload to the GCS bucket storege for the web, get gsutil and configure it with your goole account.
gsutil config
  • Create copies of the config files (any names will do). Do not modify the original config.yaml or definition-example.xml
cp config.yaml config-local.yaml
cp data/definition-example.xml definition-local.xml
  • Edit config-local.yaml: set "gleamviz_dir" and "foretold_channel" (this is non-public, sorry)
  • Make sure that directory YOUR_GLEAM_DIR/data/sims/ exists or create it

Running a batch of simulations

NOTE: Do not run any of the pipeline commands while GleamViz is running.

  • Once a day (and before first run) update the data files.
./gleambatch.py update
  • Your definition-local.xml can be updated via GleamViz import+export or just a text editor.
  • Update the settings (countries, parameters, ...) in config-local.yaml.
./gleambatch.py generate config-local.yaml definition-local.xml
  • Copy-paste the path of the created batch-file (or the suggested command).
  • Run GleamViz, all the created simulations should be there.
  • Run all simulations (there is a limit on how many at once you can run, perhaps also some daily limit?)
  • Retrieve all simulations via the blue buttons (Can be done in parallel. You do not need to export them!)
./gleambatch.py process out/batch-XXXXX/batch.yaml

  • This creates files in directory out/batch-XXXXX/

  • Push it to data-CHANNEL-gleam.json, where channel is staging (testing), main or anything else (will beavailable at URL )

./push_to_bucket.sh out/DATE-TIME-gleam.json data-staging-gleam.json

Installing and running GLEAMViz in Linux

When you install GleamVIz in Linux, it adds LD_LIBRARY_PATH="GLEAMviz/libs/" to your configuration in .bashrc. I would suggest you remove it as gleam comes with its own copies of some low-level libraries. (It caused some issues on my system due to libc conflicts.)

If you do the above, run GleamViz with LD_LIBRARY_PATH="GLEAMviz/libs/" GLEAMviz/gleamviz.

Some people installing GleamViz on Linux have the issue that GLEAMViz crashes on start (sometimes complaining about some Qt dependency, I do not remember the exact error message). The root cause I found was that GleamViz ships with a part of libc, namely its own libm.so.6 from libc 2.29 (mine is 2.30). Renaming it helped resolve it for me and Daniel.