---
title: "GWAS Tracks"
author: "Paul Shannon"
package: igvR
date: "`r Sys.Date()`"
output:
   BiocStyle::html_document
vignette: >
  %\VignetteIndexEntry{"GWAS Tracks"}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

<style>
.main-container { width: 1000px; max-width:2800px;}
</style>


```{r setup, include = FALSE}
options(width=120)
knitr::opts_chunk$set(
   collapse = TRUE,
   eval=interactive(),
   echo=TRUE,
   comment = "#>"
)
```


# Overview

Genome Wide Association Studies (GWAS) are often viewed in a "manhattan plot".
[igv.js](https://github.com/igvteam/igv.js/wiki/GWAS) supports thesein two formats: a fixed bed format, and
the traditional flexible, multi-column file format in which GWAS results are often summarized.

We support only the latter since its flexibility can be used with the five-column fixed bed format as well.


The data from GWAS studies is rather sparse, usually with no more than a few hundred variant
hotspots which associate with the phenotype of interest.  Another feature of these
data is that the community is often most concerned with the statistical significance
of the association.  This is expressed as a pvalue.  A minus-log10 transform of the pvalue
controls the height of the rendered variant in the manhattant plot.

Support for this track type in igvR is new (summer 2022) and unpolished.  Control of
autoscaling and fixed scaling needs work.

There are two relevant classes, shown here with their constructors:

1. GWASTrack("bellenguuez", tbl.gwas, chrom.col=1, pos.col=2, pval.col=5, trackHeight=80)
2. GWASUrlTrack("igv sample", url, chrom.col=12, pos.col=13, pval.col=28)

All the arguments are the same, except for the second one, which is either a local data.frame in memory,
or a url pointing to a remote file served over http.

# Demostration: a local data.frame with 5 columns

Here we use a 5-column data.frame from a file hosted in igvR's extdata, GWAS results from a 2022
Nature Genetics paper, [New insights into the genetic etiology of Alzheimer’s disease and related
dementias](https://www.nature.com/articles/s41588-022-01024-z).

Its first few rows:
```
  chrom     start       end        name    score
1  chr1 109345809 109345810 rs141749679 2.36e-04
2  chr1 207577222 207577223    rs679515 1.51e-14
3  chr2   9558881   9558882  rs72777026 4.14e-03
4  chr2  37304795  37304796  rs17020490 2.33e-04
5  chr2 105749598 105749599 rs143080277 6.29e-04
6  chr2 127135233 127135234   rs6733839 2.06e-30
```

```{r eval=FALSE}
library(igvR)
igv <- igvR()
setBrowserWindowTitle(igv, "AD GWAS")
setGenome(igv, "hg38")
tbl.gwas <- read.table(system.file(package="igvR", "extdata", "gwas", "bellenguez.bed"),
                       sep="\t", as.is=TRUE, header=TRUE, nrow=-1)
track <- GWASTrack("bellenguuez", tbl.gwas, chrom.col=1, pos.col=2, pval.col=5, trackHeight=80)
displayTrack(igv, track)
```

```{r, eval=TRUE, echo=FALSE, out.width="95%"}
knitr::include_graphics("images/gwas-01.png")
```

Zoomed into the neighborhood of TREM2, a gene independently known to be associated with Alzheimer's
Disease.

```{r, eval=TRUE, echo=FALSE, out.width="95%"}
knitr::include_graphics("images/gwas-02.png")
```


# Demonstration: a remote GWAS file with 34 columns

This file is more typical of that used in GWAS studies.  These are the columns:

  1. DATE ADDED TO CATALOG
  2. PUBMEDID
  3. FIRST AUTHOR
  4. DATE
  5. JOURNAL
  6. LINK
  7. STUDY
  8. DISEASE/TRAIT
  9. INITIAL SAMPLE SIZE
 10. REPLICATION SAMPLE SIZE
 11. REGION
 12. CHR_ID
 13. CHR_POS
 14. REPORTED GENE(S)
 15. MAPPED_GENE
 16. UPSTREAM_GENE_ID
 17. DOWNSTREAM_GENE_ID
 18. SNP_GENE_IDS
 19. UPSTREAM_GENE_DISTANCE
 20. DOWNSTREAM_GENE_DISTANCE
 21. STRONGEST SNP-RISK ALLELE
 22. SNPS
 23. MERGED
 24. SNP_ID_CURRENT
 25. CONTEXT
 26. INTERGENIC
 27. RISK ALLELE FREQUENCY
 28. P-VALUE
 29. PVALUE_MLOG
 30. P-VALUE (TEXT)
 31. OR or BETA
 32. 95% CI (TEXT)
 33. PLATFORM [SNPS PASSING QC]
 34. CNV

Here we create the track and display it:

```{r eval=FALSE}
url <- "https://s3.amazonaws.com/igv.org.demo/gwas_sample.tsv.gz"
track <- GWASUrlTrack("igv sample", url,chrom.col=12, pos.col=13, pval.col=28)
displayTrack(igv, track)
```
First, the whole genome view:

```{r, eval=TRUE, echo=FALSE, out.width="95%"}
knitr::include_graphics("images/gwas-03.png")
```
Now, zooin into on chromsome 6, and click on any variant in the display (here we
chose  **rs7767396**) the full data from the multi-column gwas file is available.

```{r, eval=TRUE, echo=FALSE, out.width="95%"}
knitr::include_graphics("images/gwas-04.png")
```

# Session Info

```{r eval=TRUE}
sessionInfo()
```