tidy up functions for conversion between antibody vs. vaccine efficacy

[yangclaraliu]

key output:

1. cut off for L, M, H
2. percentage transition for v_b and v_p

[yangclaraliu]

• visualise the spread of distributions
• by group meta-analysis diagnostics
• Take AZ and PZ out
• take BoosterPZ and MixedVac out for sanity checks
• sanity check (a. pz > everything else; b. vac + conv > vac; day_measure; c. short elapse > long elapse; d. against alpha/ beta > Omicron)

[yangclaraliu]

• check on the pooled standard deviation with Deborah
• using difference IC50, ID50, ID80, NT50
• variance (UK/ SA, what to do with them?)

[yangclaraliu]

• density figure by vaccine combo with antibody titre on the x-axes

[yangclaraliu]

• ask Deborah about AZ_SP and SP_AZ and SP_SP and AZ_AZ
• conv+PZ_PZ_PZ and conv+PZ_PZ

[yangclaraliu]

meeting planned for in the next 1-2 weeks

[yangclaraliu]

setting up thresholds for L, M, and H

[jarawees]

@yangclaraliu This code chunk is used for converting NAb fold (of vaccine to convalescent) to vaccine efficacy against infection/mild disease by: 1) Use mean and SE of NAb fold from meta-analysis to stimulate 1000 NAb fold values for each vaccine combo (assumed normal distribution), store those stimulated values in _tbl_ab_ratiorandom 2) Convert NAb fold values from 1) to vaccine efficacy using vx_efficacy function (with covariate matrix from Deborah's email), store vaccine efficacy results in _tbl_vx_eff_montecarlo

https://github.com/jarawees/covidmodel-hitap/blob/17420d293093089a2e72fc6f2ff95645fe16d140/code/NAb_c19_meta_analysis_v3/NAb_c19_meta_analysis_v4.R#L339-L390

[jarawees]

Code chunk for plotting density figure of vaccine combo with NAb fold on the x-axis (using values from _tbl_ab_ratiorandom): https://github.com/jarawees/covidmodel-hitap/blob/17420d293093089a2e72fc6f2ff95645fe16d140/code/NAb_c19_meta_analysis_v3/NAb_c19_meta_analysis_v4.R#L407-L467

[jarawees]

Code chunk for plotting vaccine efficacy stratified by low, medium, and high NAb titre levels: https://github.com/jarawees/covidmodel-hitap/blob/7c40f30d2182ab097f8f9afe77a5286414d86ede/code/NAb_c19_meta_analysis_v3/NAb_c19_meta_analysis_v4.R#L501-L523

Notes:

• I increased the number of stimulations to 10000 and 100000 (as Mark suggested), but there were overlapped between Vl, Vm, and Vh. Maybe the overlap in vaccine efficacy plot occurred because the cut-point was made based on NAb titre?
• I run the code on the other computer and found similar plots
• If the cut-points of -0.5 and 0.5 were used, the density plot would change to