When running ChIPQC bioconductor package after peak calling, should I use sorted BAM files or unsorted BAM files?

Question

I have called peaks using MACS2 and now I'm trying to run ChIPQC bioconductor package. Can someone please tell me should I use sorted BAM files or unsorted BAM files? I have created sorted BAM files and from that indexes for BAM files. When I was trying run ChIPQC earlier, I got an error. Then I figured out that is because I don't have index files. Then I created indexes after sorting BAM files.

Answer 1

You need to provide sorted bam files, ideally with an index. If you run ChIPQCsample, it calls ChIPQC:::sampleQC which iterates through the chromosomes or contigs, using the index, and the absence of the index will crash it.

We can use an example, first in linux:

wget ftp://hgdownload.cse.ucsc.edu/goldenPath/hg19/encodeDCC/wgEncodeHaibTfbs/wgEncodeHaibTfbsA549RxlchV0416101Etoh02AlnRep1.bam -O Etoh02AlnRep1.bam

samtools index ./Etoh02AlnRep1.bam
# s
samtools view -uh Etoh02AlnRep1.bam chr1 > sorted.bam
samtools index sorted.bam
# quick way to unsort
samtools sort -n -o unsorted.bam sorted.bam
# it fails to index
samtools index unsorted.bam 
[E::hts_idx_push] Unsorted positions on sequence #1: 246814554 followed by 42922635
samtools index: failed to create index for "unsorted.bam"

Then in R:

library(ChIPQC)
ChIPQCsample("unsorted.bam")
Bam file has 25 contigs
Error in value[[3L]](cond) : 
  failed to open BamFile: failed to load BAM index
  file: unsorted.bam

ChIPQCsample("sorted.bam",annotation=NULL)
Bam file has 25 contigs
Calculating coverage histogram for chr1

Calculating SSD for chr1

Calculating unique positions per strand for chr1

Calculating shift for chr1

 300 / 300
Counting reads in features for chr1

....