In tobyjohnson/gtx: Genetics ToolboX

% PGx CSR insert % Toby Johnson % r format(Sys.time(), "%d %B %Y")

suppressPackageStartupMessages(library(knitr))
opts_chunk$set(dev = c("png", "pdf"), 
               fig.width = 6, fig.height = 5, dpi = 300)
suppressPackageStartupMessages(library(gtx)) 
# the gtx package is available from https://github.com/tobyjohnson/gtx

gtxpipe.summary <- read.csv("01_study_summary.csv", as.is = TRUE, row.names = 1)
gtxpipe.groups <- read.csv("02_subject_disposition.csv", as.is = TRUE, row.names = 1)
#Cannot read first column in as row.names - not guaranteed to be unique
gtxpipe.demotable <- read.csv("03_subject_demographics.csv", as.is = TRUE, row.names = NULL, check.names = FALSE)
gtxpipe.results <- read.csv("04_summary_results.csv", as.is = TRUE) #, row.names = 1)
# in pipeline should format(gtxpipe.results$thresh1) and thresh2
gtxpipe.results$lambda <- formatC(gtxpipe.results$lambda, digits = 3, format = "f")
gtxpipe.results$thresh1 <- formatC(gtxpipe.results$thresh1, digits = 2, format = "e")
gtxpipe.results$thresh2 <- formatC(gtxpipe.results$thresh2, digits = 2, format = "e")

Objective(s)

The pharmacogenetic (PGx) study utilized data from this clinical study and genome-wide germline genetic data to test for association between genetic variants and differential efficacy response to [SPECIFY NAMES].

Summary

Of r gtxpipe.summary['All enrolled, ITT', 'value'] subjects enrolled in the clinical study, a total of r gtxpipe.summary['All enrolled, PGx', 'value'] (r gtxpipe.summary['Overall PGx percent', 'value']%) subjects provided PGx consent, a sample for genetic analysis and were successfully genotyped. For PGx analyses, r gtxpipe.summary['PGx group overlap', 'value'] treatment group(s) were defined as listed in Table 1. Genotype-by-treatment interaction effects were estimated using contrasts between main effects estimated independently for each treatment group.

Table 1. Disposition of subjects by PGx treatment group and availability of PGx data.

kable(subset(gtxpipe.groups, group %in% gtxpipe.results$group,
             select = c("group", "arms", "N.ITT", "N.PGx")), 
      row.names = FALSE)

Variables used in PGx analyses are summarized in Table 2. For each variable, the distribution in all ITT subjects, and in the PGx sample, are compared, by treatment group.

Table 2. Demographics of subjects included in PGx analyses.

kable(gtxpipe.demotable)

[VARIABLE] was inverse normal transformed prior to association analysis. ** [VARIABLE] was derived by [insert brief summary]. [IF NECESSARY COMMENT ON ANY GROSS IMBALANCES (see hidden text above for guidance)]

PGx analyses conducted are summarized in Table 3. The specific candidate genetic variants (CG) analysed are listed in Appendix Table 1. The number of significantly associated variants are summarised in total for each analysis.

Table 3. PGx analysis summary

kable(subset(gtxpipe.results, primary,
             select = c("model", "group", "lambda", "thresh1", "hits1", "thresh2", "hits2")),
      row.names = FALSE)

All analyses (whether with or without baseline clinical covariates) were adjusted for [number of] ancestry principal components to control for confounding by population structure (Price et al. 2006). [If needed:] All analyses for Treatment Group 3 were additionally adjusted for study arm. ** The genomic control coefficient λ, estimated from the genome-wide association scan (GWAS) results for each analysis/covariate/treatment group combination, provides a measure of test statistic calibration after covariate adjustment (expected value λ=1), and any residual test statistic inflation was corrected using the genomic control method (Devlin et al. 2001).

Conclusions

None of the genetic variants investigated (neither GWAS nor candidate genetic variants), were significantly associated with any of the [EFFICACY and/or SAFETY] endpoint(s) investigated.