One of the reasons to convert bam to BigWig is to get a normalized coverage in viewers such as
IGB.
Tools needed
samtools (if your bam file is not sorted by position),
bedtools and ucsc's bedGraphToBigWig in the
ucsc command line utilities
Procedure
samtools sort input.bam input.sorted
mysql --user=genome --host=genome-mysql.cse.ucsc.edu -A -e "select chrom, size from hg19.chromInfo" > hg19.genome
bedtools genomecov -ibam input.sorted.bam -bg hg19.genome > input.sorted.bedGraph
bedGraphToBigWig input.sorted.bedGraph hg19.genome input.sorted.bigwig
Unfortunately we can't pipe genomecoverage directly into bedGraphToBigWig because bedGraphToBigWig does a few passes over the file (for improved speed).
Other things to consider:
1) Scaling : setting a scale factor (within bedtools) would normalize your bedgraph/bigwig files (the default=1, no scaling). I generally scale all files to say, 10 million reads (get total reads from eg. samtools flagstat output) so that they can be consistently used throughout different comparisons.
2) -split option should be used for RNA-seq to properly represent spliced reads.
3) -strand option in theory would give you strand-specific files (meaning you'll have a +ve and -ve set) but for my paired-end directional reads, it seems to only work for the first in read pair. I've left out this option until I figure out a solution.
