-
Notifications
You must be signed in to change notification settings - Fork 2
/
IntroductionCoPheScan_01.Rmd
84 lines (55 loc) · 3.23 KB
/
IntroductionCoPheScan_01.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
---
title: "Introduction to CoPheScan"
author: "Ichcha Manipur"
date: "`r Sys.Date()`"
output: rmarkdown::html_vignette
vignette: >
%\VignetteIndexEntry{Introduction to CoPheScan}
%\VignetteEngine{knitr::rmarkdown}
%\VignetteEncoding{UTF-8}
---
```{r, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "Intro-"
)
```
```{r setup, message=FALSE, warning=FALSE}
library(cophescan)
```
#### CoPheScan
The cophescan package implements Coloc adapted Phenome-wide Scan (CoPheScan), a Bayesian method to perform Phenome-wide association studies (PheWAS) that identifies causal associations between genetic variants and phenotypes while simultaneously accounting for confounding due to linkage disequilibrium.
Given a query variant and genomic region with Q SNPs for a query trait, cophescan discriminates between 3 hypotheses:
$H_n$ : No association with the query trait (1 configuration)
$H_a$ : Association of a variant other than the query variant with the query trait (Q-1 configurations)
$H_c$ : Association of the query variant with the query trait (1 configuration)
with $p_n$, $p_a$ and $p_c$ being their corresponding priors.
#### CoPheScan approaches
cophescan can be run in different ways depending on the size and type of dataset.
First, choosing the method for Bayes factor computation:
| | Single causal variant | Multiple causal variants | Requires LD matrix |
|-------|-----------------------|--------------------------|--------------------|
| ABF | ✔ | x | No |
| SuSIE | ✔ | ✔ | Yes |
| | | | |
Whenever, LD matrices are available (preferably in-sample LD), \``cophe.susie`\` is the recommended method as it accounts for multiple causal variants in the tested region.
Next, depending upon the size of the dataset we choose the method to specify priors :
| | Dataset | Inclusion of covariates |
|---------------------|---------|-------------------------|
| Fixed priors | Small | \- |
| Hierarchical priors | Large | ✔ |
| | | |
The different combinations that can be run are:
ABF/Fixed priors: `cophe.single`
SuSIE BF/Fixed priors: `cophe.susie`
ABF/Hierarchical priors: `cophe.single.lbf` + `run_metrop_priors`
SuSIE BF/Hierarchical priors: `cophe.susie.lbf` + `run_metrop_priors`
#### Further reading
1. Description of the CoPheScan method:
[CoPheScan: phenome-wide association studies accounting for linkage disequilibrium](https://doi.org/10.1101/2023.06.29.546856)
2. coloc: [Giambartolomei et al (2013)](https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1004383)
3. coloc with SuSIE: [Wallace et al (2021)](https://doi.org/10.1371/journal.pgen.1009440), [github](https://github.com/chr1swallace/coloc)
4. ABF: [Wakefield (2008)](https://doi.org/10.1002/gepi.20359)
5. SuSIE: [Wang et al (2020)](https://doi.org/10.1111/rssb.12388), [github](https://github.com/stephenslab/susieR)
------------------------------------------------------------------------