master_document_with_validation.Rmd

---
title: 'Supplemental Statistical Analysis for "Exploring the function of the BBSome using clinical data: Meta-analysis of genotype-phenotype associations in Bardet-Biedl Syndrome"'
output:
  pdf_document: 
     toc: true
---


***

This document consists of four parts: 

* Part 1 describes the Bayesian analysis reported in the main manuscript.
* Part 2 describes all Bayesian models we tried throughout this project and discusses the reasoning behind our choice of model for the main analysis, in particular why between-study variability is crucial and taking into account complete loss of function (cLOF) useful while age, and sex can be omitted.
* Part 3 discusses the overall ability of the model to predict a validation dataset
* Part 4 shows how the conclusions of the paper hold under multiple different models and in the validation data.

The complete source code for the analysis can be found at https://github.com/martinmodrak/bbs-metaanalysis-bayes or Zenodo, DOI: 10.5281/zenodo.3243264

```{r setup, echo=FALSE, message = FALSE, warning=FALSE}
knitr::opts_chunk$set(echo=FALSE, cache = TRUE, cache.lazy = FALSE)
```

\newpage

# Part 1: Main Bayesian Analysis

```{r child_main, child = 'main_analysis.Rmd'}
```

\newpage

# Part 2: Alternative Models & Model Selection

```{r child_alternative, child = 'alternative_models.Rmd'}
```

\newpage

# Part 3: Validation on a new dataset

```{r child_validation, child = 'validation_dataset.Rmd'}
```

\newpage

# Part 4: Conclusions under Multiverse Analysis and Validation

```{r child_multiverse, child = 'multiverse_analysis.Rmd'}
```


\newpage

# Original computing environment

```{r, cache=FALSE}
git_rev <- tryCatch({system("git rev-parse HEAD", intern=TRUE)}, error = function(e) {"Could not read Git revision"})
```

This report was built from Git revision ``r git_rev`` on `r format(Sys.time(), "%d %B, %Y")`

```{r}
sessionInfo()
```