bedGraphToBigWig does not work with single base coverage

[GenoMax]

If you are going to use UCSC browser to display the files consider this: http://genome.ucsc.edu/FAQ/FAQtracks.html#tracks1

[biocomputer]

So you just have to add 1 to each value in the end position column? I'm still learning this stuff so there's probably a better way to do this, but this command below worked when I tested it on a small file. It assumes tab-separated columns, should add 1 to each value in the 3rd column, and should print out any non-position lines (ie. any lines that don't start with "chr" like a file header) as is. Test first and use at your own risk

awk '/^chr/ {FS="\t";$3=$3+1;print} !/^chr/{print}' inputFile > outputFile

Or instead of adding 1 to the end position do you need to substract 1 from each start position?

awk '/^chr/ {FS="\t";$2=$2-1;print} !/^chr/{print}' inputFile > outputFile

Here's another version that should do the same thing, maybe it'll be faster?

awk 'BEGIN { OFS="\t"}{if (/^chr/) $3=$3+1; print}' inputFile > outputFile

[blancha]

Great. Thanks BioComputer.
That is exactly the formal definition I was looking for.

I did waste half a day figuring out the "bug", though.
Thankfully, you had a quick answer to my question, as I would never have figured out the reason behind the "bug" on my own.
I guess it's more of a "feature".
Now, I just need to decide if I should write a Python script to correct the coordinates, or whether if I run bismark_methylation_extractor again with the --zero_based parameter I will get the desired format.

I used bismark_methylation_extractor to generate the BED files. The output was a bit confusing since, by default, the bedgraph file outputted has 0-based start coords and 1-based end coords, whereas the other BED files generated have 1-based genomic coords.

To be fair, it is in the documentation, although I'm still unsure if adding the --zero-based parameter will result in 0-based start and end coordinates, or just 0-based start coordinates, and 1-based end coordinates. I imagine it is the latter.
Bioinformatics is so much fun.

Thanks again!

[biocomputer]

I think your bedgraph should be the second way, 1 2 not 1 1, because bedgraph files are "half open".

[blancha]

Hi,

I've identified a "bug" in UCSC's bedGraphToBigWig.
It does not work with single base coverage.

If I have a file where the start and end are always equal (i.e. single base coverage), the bigWig file generated by bedGraphToBigWig is empty.
Does anyone have a workaround, or am I simply not respecting the format for bedgraph files? If I want to include the count just for base 1, how am I supposed to write it?
1 1 count
or
1 2 count
The 1st option appears correct but bedGraphToBigWig will not take it.
It would be a big job to convert all my bedgraph files to the format bedGraphToBigWig appears to want.

Here are 2 example bedgraph files: test1.bedgraph with the start and the end equal to 1, and test2.bedgraph where the start and end are 1 and 2, respectively. The bigWig files generated by bedGraphToBigWig is empty in the first case, and with 1 base covered in the second case.

blancha@sw-2r12-n36 ~]$ more test1.bedgraph
chr1 1 1 1
[blancha@sw-2r12-n36 ~]$ more test2.bedgraph
chr1 1 2 1
[blancha@sw-2r12-n36 ~]$ bedGraphToBigWig test1.bedgraph UCSCChromInfo.txt test1.bw
[blancha@sw-2r12-n36 ~]$ bedGraphToBigWig test2.bedgraph UCSCChromInfo.txt test2.bw
[blancha@sw-2r12-n36 ~]$ bigWigInfo test1.bw
version: 4
isCompressed: yes
isSwapped: 0
primaryDataSize: 30
primaryIndexSize: 6,204
zoomLevels: 1
chromCount: 1
basesCovered: 0
mean: -nan
min: 1.000000
max: 1.000000
std: -nan
[blancha@sw-2r12-n36 ~]$ bigWigInfo test2.bw
version: 4
isCompressed: yes
isSwapped: 0
primaryDataSize: 32
primaryIndexSize: 6,204
zoomLevels: 1
chromCount: 1
basesCovered: 1
mean: 1.000000
min: 1.000000
max: 1.000000
std: 0.000000

Thank you for your help.