computeMatrix scale-regions A region that is shorter than the bin size

[sunta3iouxos]

When I am running the following command:

computeMatrix scale-regions --regionsFileName assemblies/mm39/annotation/genes.gtf --scoreFileName $bw --outFileName H3_genebody_ncbiRefSeq_mm39_scaled_noSpikeregions_unscaled.gz --outFileNameMatrix H3_genebody_scaled_noSpikeregions_unscaled.tab --outFileSortedRegions H3_genebody_scaled_noSpikeregions_unscaled.bed --upstream 2000 --downstream 2000 --blackListFileName assemblies/mm39/annotation/blacklist.bed --smartLabels -p 13 --missingDataAsZero --skipZeros

I only get some warning that some genes in the gene list are not represented in the bw files. Which is perfectly understandable.

But when I am adding the unscaled options I got the following, even if I use the 3prime or the 5prime option, or both:

computeMatrix scale-regions --regionsFileName /mnt/Z/Theodoros\ Georgomanolis/assemblies/mm39/annotation/genes.gtf --scoreFileName $bw --outFileName H3_genebody_ncbiRefSeq_mm39_scaled_noSpikeregions_unscaled.gz --outFileNameMatrix H3_genebody_scaled_noSpikeregions_unscaled.tab --outFileSortedRegions H3_genebody_scaled_noSpikeregions_unscaled.bed --upstream 2000 --downstream 2000 --blackListFileName /mnt/Z/Theodoros\ Georgomanolis/assemblies/mm39/annotation/blacklist.bed --smartLabels -p 13 --missingDataAsZero --skipZeros --unscaled5prime 1000 --unscaled3prime 1000

A region that is shorter than the bin size (possibly only after accounting for unscaled regions) was found: (-3920) ENSMUST00000083636.3 chr12:109709722:109709802. Skipping...

Also it appears that this happens for all genes. The bs is the default one, and was the same while I produced the bw with the bamCoverage commaND:

bamCoverage -b ${bam}.bam -o ${bw}_scaled.bw --scaleFactor $c -p 8 --ignoreForNormalization $normalisation –bs 10 --extendReads 150 --centerReads --normalizeUsing None --effectiveGenomeSize 2654621783

Cound the extendReads be the issue?

[WardDeb]

note that computeMatrix also has a binSize argument to, which defaults to 10. in your case, the 'region shorter then the binsize, after accounting for unscaled regions' refers to: length of region - (unscaled5prime + unscaled3prime) = region - 2000, since this number is smaller than 10 the region is skipped.