PvMolCorrection

Plasmodium vivax molecular correction using Pv3Rs: a re-analysis of P. vivax genetic data from the Vivax History (VHX) trial (Chu et al. Clin. Infect. Dis. 67 (2018)) and the Best Primaquine Dose trial (Chu et al. Clin. Infect. Dis. 68 (2018)). The original analysis on which this re-analysis is based is described in Taylor & Watson et al. Nat. Commun. 10 (2019). The re-analysis uses original input and output in jwatowatson-RecurrentVivax-4870715, which was downloaded from zenodo.org/records/3368828, and from MS_Final, which was saved as MS_final_generated_by_running_all_chunks_of_Pooled_Analysis.Rmd in RData/ after running Pooled_Analysis.Rmd in jwatowatson-RecurrentVivax-4870715/Pooled_Final_Analysis/.

Scripts required to re-produce the analysis

1. Run Compute_recurrent_states_Pv3Rs.R: takes several hours and generates

   • prior_estimates.RData
   • results_Pv3Rs.RData

2. Run Analysable_data.R to generate

   • tab_MOIs.csv (used to create tab_MOIs_edited.png)
   • data_repeat_episode.png
   • data_uncomputable.png

3. Run Extract_and_plot_Pv3Rs.R to generate

   • marg_results_Pv3Rs.RData
   • simplex_Pv3Rs.png

4. Run Extract_and_plot_prototype.R to generate

   • simplex_prototype.png
   • compare_Pv3Rs_vs_prototype.png
   • data_Pv3Rs_vs_prototype.png

5. Run Compute_FDR_Pv3Rs.R to generate false discovery rates and

   • RData/ys_null.RData
   • RData/probs_null.RData
   • RData/min_probs_null.RData

6. Run Compare_joint_pwise_Pv3Rs.R to generate

   • compare_joint_vs_pwise.png
   • data_joint_vs_pwise.png

7. Run Compute_null.R to generate

   • RData/nulls.RData
   • RData/quantiles_nulls.RData

8. Run Compute_genetic_proximity.R to generate

   • RData/genetic_proximities.RData

9. Run Plot_genetic_proximity.R to

   • decide on genetic distance
   • check superiority of Pv3Rs to the prototype
   • Generate probability of failure among reinfection-classified
   • Generate probability of reinfection among failure-classified
   • Generate Figures/compare_prob_prox.pdf
   • Generate Figures/outlier_correction.pdf
   • Generate data_outliers.png

10. Run Compute_failure_rates.R to generate

    • compare_failure_rates.png

Extra scripts

• To better understand data input and estimates generated by the prototype (can be skipped), run the Understanding_*.R scripts, noting that Understanding_priors.R requires prior_estimates.RData to be first generated by running Compute_recurrent_states_Pv3Rs.R.

In no particular order:

• Cherry_picked_examples.R generates cherries.png.
• Plot_site_freq_spectrum.R to plot per-marker SFS ordered by marker cardinality. Nothing notable to report: SFS are remarkable similar.
• Check_prior_conversion.R to assess the accuracy of prior conversion given access to only marginal probabilities.
• Check_markers_typed.R to check the assertion that additional 6 markers typed if matches seen a > 1/3 first three markers.
• Compare_failure_recur_independence.R to see the impact of computing probabilities of failure assuming and not independence between recurrences
• Visualise_follow_up.R to see how first recurrences recover follow up protocol
• Chance_matching.R to see how probabilities assuming iid draws compare with Dcifer relatedness estimates and Pv3Rs posteriors.