Week 22/07/2024

[olivierlabayle]

Thesis Writing:

• [ ] The not so close closest estimands
• [ ] TarGene discussion
• [ ] Describe available workflows
• [x] Background: OneStep and TMLE

Simulation Study:

• [ ] Migrate the reusable code to Simulations.jl and further test
• [ ] High dimensional variance should include the identity matrix
• [ ] Null Simulation Analysis
  • [x] Run (Resume once realistic is done)
  • [x] Check if all linked variants are in HLA region
  • [ ] Generate same plots
• [ ] Realistic Simulation Analysis
  • [ ] Resume on crash (Watchout the commit!! Save first!)
  • [ ] Analyse when run finished and ad to thesis

DB (Lead Breeshey):

• [ ] Review PR

VDR paper:

• [ ] Setup Joint Repo for Kelsey and I
• [ ] Write joint config
• [ ] Rerun both
• [ ] Analyse

TarGene CleanUp:

• [ ] https://github.com/TARGENE/targene-pipeline/issues/179#issue-2303134191
• [x] Update and Add more native estimator configs
• [ ] MultiArch docker images across org?
• [ ] executable TMLE CLI
• [x] estimator configs by name at targene level
• [x] Remove the STUDY_DESIGN parameter and infer the mode(s) from the yaml/jls file(s)?
• [ ] Release Targene packages in julia registry for better deps management
• [ ] Add a Quick Start section
• [x] In the aggregation of results, make a single vector of results in the hdf5 which will likely improve performance
• [x] Make that by default targeted estimation only saves once (without the influence curves)
• [x] Maybe change default caching strat to max size?
• [ ] Add a "famous" simulation to TMLE.jl docs?
• [x] drop influence curve straight after estimation and regardless of output
• [x] Add defaultdouble resources per task attempt for all processes
• [x] Add a targene flat config test