Four common classes of annotation

These are the objects with the transcriptome information

• Saved using GRanges classes
• Derived heavily from the GenomicRanges & IRanges packages
• The key idea is to refer the the genome using ranges to define locations

library(BiocInstaller)
biocLite("TxDb.Hsapiens.UCSC.hg19.knownGene")

library(TxDb.Hsapiens.UCSC.hg19.knownGene)
txdb <- TxDb.Hsapiens.UCSC.hg19.knownGene   

txdb

This will load all the package dependencies as well

Let's look at a GRanges object

Note that our txdb object used EntrezGene Ids

ids <- c(BRCA1="672", PTEN="5728")
genes(txdb, filter=list(gene_id=ids)) 

Run Length Encoding format vectors

• More memory efficient way to store positional information
• highly efficient for long regions of "no information", or
• also efficient for data with long stretches of repeats

rle : Part of the base R package

Rle : S4Vectors version

Rle is used extensively in GenomicRanges

x <- c(1, 0, 0, 0, 1, 1, 2, 0, 0)
Rle(x)

gr <- GRanges(seqnames=Rle(c("chr1", "chrMT"), c(2, 4)),
              ranges=IRanges(15:20, 20),
              strand=rep(c("+", "-", "*"), 2))

Print the object by typing gr

The essential components are:

• seqnames & ranges
• If strand is omitted, the value * is added

Try these commands:

seqnames(gr)
strand(gr)
ranges(gr)
seqinfo(gr)
length(gr)
gr[1]
width(gr)
start(gr)

• seqinfo() returns an object with a formal class
• Seqinfo objects contain metadata about each sequence

names(gr) <- paste0("Rng", LETTERS[1:length(gr)])

We can assign names to the ranges:

• Could be exons, genes, SNPs, CDS or any other feature

Now look at the object again

We can also add some key information about the sequences

seqlengths(gr) <- c(5e6, 1.5e5)
isCircular(gr) <- c(FALSE, TRUE)
genome(gr) <- c("madeUp.v1")
seqinfo(gr)

GRanges objects also have columns for metadata

Let's add:

1. Some \(p\)-values from a hypothesis test
2. Alternative names for the Chromosomes

mcols(gr) <- data.frame(score = 10^(-rexp(6)),
                        altChr = rep(c("G001", "G002"), 
                                     times=c(2, 4)))

Try these commands:

gr[1:3]
gr[1:2, 1]
subset(gr, score < 0.05)
subset(gr, width==1)
subset(gr, start > 18)
subset(gr, start > 18 | width ==5)
table(gr$altChr)
summary(mcols(gr)[,"score"])

GRanges objects can also be extended to GRangesList objects

exByGn <- exonsBy(txdb, "gene")
length(exByGn)

exByGn

As well as the exonsBy() methods, other methods include

• transcriptsBy(), cdsBy(), threeUTRsByTranscript() + more

In the current example exons are listed by gene, but can also be listed by exon, cds or tx

These behave like normal list objects in R

Try these commands

exByGn[[1]]
exByGn$`1`
exByGn[1:2]
sapply(exByGn[1:10], 
       function(x){length(subset(x, width<100))}) 
unlist(exByGn[1:5])

Ask if you're unsure about what any of the above commands do

Also contains other useful methods

promoters(txdb, upstream=100, downstream=50,
          columns = c("tx_name", "gene_id"))

library(mirbase.db)
microRNAs(txdb)[1:3]

In our object exByGn, we have a list of genes and their ranges

• Named using EntrezGene IDs
• Let's get EnsemblIDs and the biotype
• Setup a biomaRt connection

library(biomaRt)
mart <- useMart(biomart = "ENSEMBL_MART_ENSEMBL", 
                dataset = "hsapiens_gene_ensembl")

Now we collect the genes and define the attributes we want

entrezIDs <- names(exByGn)
attr <- c("entrezgene", "ensembl_gene_id", "external_gene_name", "gene_biotype")
results <- getBM(attributes = attr,
                 filters = "entrezgene",
                 values = entrezIDs,
                 mart= mart)

How did we go?

summary(entrezIDs %in% results$entrezgene)
table(table(results$entrezgene))

Let's just keep the first mapping

library(dplyr)
results <- results %>%
  distinct(entrezgene, .keep_all = TRUE) %>%
  mutate(entrezgene = as.character(entrezgene))

We also might need to convert those IDs back to characters

Now we can just use the function left_join() from dplyr

merged <- data_frame(entrezgene = entrezIDs) %>%
  left_join(results)