Skip to content

Causal inference, super learning and credible interval generation code for analyzing P. falciparum drug resistance genotypes

Notifications You must be signed in to change notification settings

breesie/percolateR

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

This repository contains two scripts related to the study of molecular markers of antimalarial drug resistance

  1. Beta_binomial_percolateR is iterations of a function which take prior genotyping data in the form of prevalence and current genotyping data generated via recent benchwork

Prior information containing count data of both mutant infections, and sample size, were coded via a beta distribution.

Π ~ Beta(α,β) α = Prior mixed + Prior mutant infections β = Prior WT infections

Our observed outcome (prevalence) is distributed binomially, generating the following likelihood function.

Y ~ Binomial(Observed mutant infections, Π)

The posterior distribution of prevalence given our observed data is described via a beta-binomial conjugate prior. P(Π|Y) = Beta(α^1,β^1) α^1 = Prior mutant infection + observed mutant infections β^1 = Prior WT infections + N observed - observed mutant infections

All years which had no corresponding information are parameterized via Jeffreys prior, an uninformative prior of the beta distribution. α uninformative = .5, β uninformative = .5

unnamed

  1. k13_TMLE is an implementation of a "causal inference estimator" targeted maximum likelihood estimation for the DAG below, done in order to understand the unidentified impact of indoor residual spraying on the emergence of k13 mutations within Ugandan Plasmodium falciparum populations. This implementation respects district level clustering. This code is annotated for reproducibility, and has been used to implement TMLE in other studies.

unnamed

Our wished for causal parameter is the mean difference in probability of contracting a P.f. infection with a WHO validated/candidate k13 polymorphism if, contrary to the fact, every study subject lived in district received IRS, vs if, contrary to the fact, no study subject lived in district received IRS.

Ψ*(P*) = E*(Y1) - E*(Y0)

About

Causal inference, super learning and credible interval generation code for analyzing P. falciparum drug resistance genotypes

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published