Sprint 20/21

[cristinaetrv]

Due Nov 28/Dec 19

GRIP presentation is Dec 17
Slides will be reviewed Dec 10 (11am)
Potential topics:

• Basic LLM demo if possible
• Get data back from protein analysis (EJ, ED)
• Imputation
• Ancestry, new model, missingness
• PRS method to highlight
• Collab with Dave PRS paper (Nat Genetics?)

Imputation

• hg38 support

PRS

• Allow harmonized GWAS entered by user
• Research PRS + rare variants

Ancestry

• Writeup of performance old vs new, missingness approach
• Check if unrelated set of 1kgp changes performance in old model (minus kids from trios)

Fair PCA

• Plan out additional data in paper

• Thomas's microRNA data

• Talk to Taylor about UKBB collab

• Talk to Dave, Mark, Mike B about using data for this paper

• Additional notebooks to demonstrate

• Determine top impact journal to send to

• Test/train splits for CI for AUC

• Determine limits on balance of sensitive variables (another figure?)

• Test unrelated 1000G chr21 to see if that affects clusters

• Test Element wise PCA on 1kgp