Permalink
Switch branches/tags
Nothing to show
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
296 lines (222 sloc) 8.8 KB
---
title: Annotating Genomic Variants
author: Valerie Obenchain
output:
rmdformats::readthedown:
toc_depth: 3
number_sections: true
---
# Background
The VariantAnnotation package has facilities for reading in all or subsets
of Variant Call Format (VCF) files. These text files contain meta-information
lines, a header line and data lines each containing information about a
position in the genome. The format also may also contain genotype information
on samples for each position. More on the file format can be found in the
[VCF specs](http://samtools.github.io/hts-specs/VCFv4.2.pdf).
The 'locateVariants' function in the VariantAnnotation package identifies
where a variant is located with respect to the gene model (e.g., exon, intron,
splice site, etc.). The 'predictCoding' function reports the amino acid
change for non-synonymous coding variants. Consequences of the coding changes
can be investigated with the SIFT and PolyPhen database packages. We'll use
these functions to learn about variants located on the TRPV gene on
chromosome
# Set Up
This workflow requires several different Bioconductor packages. Usage of each
will be described in detail in the following sections.
```{r, echo=FALSE}
suppressPackageStartupMessages(library(VariantAnnotation))
suppressMessages(suppressPackageStartupMessages(library(cgdv17)))
suppressPackageStartupMessages(library(org.Hs.eg.db))
suppressPackageStartupMessages(library(TxDb.Hsapiens.UCSC.hg19.knownGene))
suppressPackageStartupMessages(library(BSgenome.Hsapiens.UCSC.hg19))
suppressPackageStartupMessages(library(PolyPhen.Hsapiens.dbSNP131))
```
```{r, eval=FALSE}
library(VariantAnnotation)
library(cgdv17)
library(org.Hs.eg.db)
library(TxDb.Hsapiens.UCSC.hg19.knownGene)
library(BSgenome.Hsapiens.UCSC.hg19)
library(PolyPhen.Hsapiens.dbSNP131)
```
Use biocLite() to get the packages you don't have installed:
```{r, eval=FALSE}
## try http:// if https:// URLs are not supported
source("https://bioconductor.org/biocLite.R")
biocLite("mypackage")
```
# Exploring variants in the TRPV gene family
This workflow focuses on variants located in the Transient Receptor
Potential Vanilloid (TRPV) gene family on chromosome 17. Sample data are
from the Bioconductor cgdv17 experimental data package which contains
Complete Genomics Diversity panel data for chromosome 17 on 46 individuals.
For more background on how these data were curated see the package vignette.
```{r, eval=FALSE}
browseVignettes("cgdv17")
```
We use a VCF file from the package that is a subset of chromosome 17 for a
single individual from the CEU population.
```{r}
library(VariantAnnotation)
library(cgdv17)
file <- system.file("vcf", "NA06985_17.vcf.gz", package = "cgdv17")
```
## Examine header data in a vcf file
To get an idea of what data are in the file we look at the header.
scanVcfHeader() parses the file header into a VCFHeader object and the
info() and geno() accessors extract field-specific data.
```{r}
hdr <- scanVcfHeader(file)
info(hdr)
geno(hdr)
```
Variants in the VCF have been aligned to NCBI genome build GRCh37:
```{r}
meta(hdr)$META
```
<p class="back_to_top">[ <a href="#top">Back to top</a> ]</p>
## Convert gene symbols to gene ids
Use the org.Hs.eg.db package to convert gene symbols to gene ids.
```{r}
## get entrez ids from gene symbols
library(org.Hs.eg.db)
genesym <- c("TRPV1", "TRPV2", "TRPV3")
geneid <- select(org.Hs.eg.db, keys=genesym, keytype="SYMBOL",
columns="ENTREZID")
geneid
```
<p class="back_to_top">[ <a href="#top">Back to top</a> ]</p>
## Create gene ranges
We use the hg19 known gene track from UCSC to identify the TRPV gene ranges.
These ranges will eventually be used to extract variants from a regions in
the VCF file.
Load the annotation package.
```{r}
library(TxDb.Hsapiens.UCSC.hg19.knownGene)
txdb <- TxDb.Hsapiens.UCSC.hg19.knownGene
txdb
```
Our VCF file was aligned to a genome from NCBI and the known gene track was
from UCSC. These institutions have different naming conventions for the
chromosomes. In order to use these two pieces of data in a matching
or overlap operation the chromosome names (also called sesqlevels) need to
match. We will modify the txdb to match the VCF file.
```{r}
txdb <- renameSeqlevels(txdb, gsub("chr", "", seqlevels(txdb)))
txdb <- keepSeqlevels(txdb, "17")
```
Create a list of transcripts by gene:
```{r}
txbygene = transcriptsBy(txdb, "gene")
```
Create the gene ranges for the TRPV genes
```{r}
gnrng <- unlist(range(txbygene[geneid$ENTREZID]), use.names=FALSE)
names(gnrng) <- geneid$SYMBOL
```
<p class="back_to_top">[ <a href="#top">Back to top</a> ]</p>
## Extract variant subsets
A ScanVcfParam object is used to retrieve data subsets. This object
can specify genomic coordinates (ranges) or individual VCF elements.
Extractions of ranges (vs fields) requires a tabix index.
See ?indexTabix for details.
```{r}
param <- ScanVcfParam(which = gnrng, info = "DP", geno = c("GT", "cPd"))
param
## Extract the TRPV ranges from the VCF file
vcf <- readVcf(file, "hg19", param)
## Inspect the VCF object with the 'fixed', 'info' and 'geno' accessors
vcf
head(fixed(vcf))
geno(vcf)
```
<p class="back_to_top">[ <a href="#top">Back to top</a> ]</p>
## Variant location in the gene model
The locateVariants function identifies where a variant falls with respect to
gene structure, e.g., exon, utr, splice site, etc. We use the gene model
from the TxDb.Hsapiens.UCSC.hg19.knownGene package loaded eariler.
```{r, eval=FALSE}
## Use the 'region' argument to define the region
## of interest. See ?locateVariants for details.
cds <- locateVariants(vcf, txdb, CodingVariants())
five <- locateVariants(vcf, txdb, FiveUTRVariants())
splice <- locateVariants(vcf, txdb, SpliceSiteVariants())
intron <- locateVariants(vcf, txdb, IntronVariants())
```
```{r}
all <- locateVariants(vcf, txdb, AllVariants())
```
Each row in cds represents a variant-transcript match so multiple rows
per variant are possible. If we are interested in gene-centric questions
the data can be summarized by gene regardless of transcript.
```{r}
## Did any variants match more than one gene?
table(sapply(split(mcols(all)$GENEID, mcols(all)$QUERYID),
function(x) length(unique(x)) > 1))
## Summarize the number of variants by gene:
idx <- sapply(split(mcols(all)$QUERYID, mcols(all)$GENEID), unique)
sapply(idx, length)
## Summarize variant location by gene:
sapply(names(idx),
function(nm) {
d <- all[mcols(all)$GENEID %in% nm, c("QUERYID", "LOCATION")]
table(mcols(d)$LOCATION[duplicated(d) == FALSE])
})
```
<p class="back_to_top">[ <a href="#top">Back to top</a> ]</p>
## Amino acid coding changes in non-synonymous variants
Amino acid coding for non-synonymous variants can be computed
with the function predictCoding. The BSgenome.Hsapiens.UCSC.hg19
package is used as the source of the reference alleles. Variant
alleles are provided by the user.
```{r}
library(BSgenome.Hsapiens.UCSC.hg19)
seqlevelsStyle(vcf) <- "UCSC"
seqlevelsStyle(txdb) <- "UCSC"
aa <- predictCoding(vcf, txdb, Hsapiens)
```
predictCoding returns results for coding variants only. As with
locateVariants, the output has one row per variant-transcript match
so multiple rows per variant are possible.
```{r}
## Did any variants match more than one gene?
table(sapply(split(mcols(aa)$GENEID, mcols(aa)$QUERYID),
function(x) length(unique(x)) > 1))
## Summarize the number of variants by gene:
idx <- sapply(split(mcols(aa)$QUERYID, mcols(aa)$GENEID, drop=TRUE), unique)
sapply(idx, length)
## Summarize variant consequence by gene:
sapply(names(idx),
function(nm) {
d <- aa[mcols(aa)$GENEID %in% nm, c("QUERYID","CONSEQUENCE")]
table(mcols(d)$CONSEQUENCE[duplicated(d) == FALSE])
})
```
The variants 'not translated' are explained by the warnings thrown when
predictCoding was called. Variants that have a missing varAllele or have an
'N' in the varAllele are not translated. If the varAllele substitution had
resulted in a frameshift the consequence would be 'frameshift'. See
?predictCoding for details.
<p class="back_to_top">[ <a href="#top">Back to top</a> ]</p>
# Exploring Package Content
Packages have extensive help pages, and include vignettes highlighting
common use cases. The help pages and vignettes are available from
within R. After loading a package, use syntax like
help(package="VariantAnnotation")
?predictCoding
to obtain an overview of help on the `VariantAnnotation` package, and
the `predictCoding` function. View the package vignette with
```{r eval=FALSE}
browseVignettes(package="VariantAnnotation")
```
To view vignettes providing a more comprehensive introduction to
package functionality use
```{r eval=FALSE}
help.start()
```
<p class="back_to_top">[ <a href="#top">Back to top</a> ]</p>
# sessionInfo()
```{r}
sessionInfo()
```
<p class="back_to_top">[ <a href="#top">Back to top</a> ]</p>