Hi all,

I am trying to analyze the Zebrafish genome and especially I want to get the proportion of the genome that is overlapped by genes.

For that purpose, I load the gtf annotation of Ensembl into "features" and then I select for genes as follows and convert to a pyrange object:

genes_=features[features['Feature']=='gene']

genes=pr.PyRanges(genes_)

Then I want to compute the length of genome overlapped by genes, as follows:

genes_lengths=genes.merge(strand).length

where strand=False when I consider both strands of the genome together (=it merges genes regardless of if they overlap on the same or opposite strand) [case1] and in [case2] I set strand to True in order to only merge genes if they overlap on the same strand.

In case1, I set genome_length to 1373471384 bp (according to Ensembl) and in case2 I set genome_length=2*1373471384 bp as I consider separately the strands.

Then I simply proceed with:

intergenic_length=genome_length-genes_lengths

percent_genic=round(get_percentage(genes_lengths,genome_length),2)

percent_intergenic=round(get_percentage(intergenic_length,genome_length),2)

In case1 I get that 55% of the genome is overlapped by genes, while in case2 I get that 28% of genome is overlapped by genes.

In both cases, it is way higher than what Ensembl reports (according to AI because I don't find it on Ensembl directly: 5 to 10% only of gene coverage).

So my question is: why am I getting such higher gene coverage and what do you think is actually correct?

Let me add that I created those two cases scenarios because I want to compare to the proportion of the genome covered by a given binding site and in case1 I would consider that when the protein binds on one strand it modify the opposite strand at the same time, while in case2 the protein can only modify the binding site on the strand on which it's bound.

Thank you deeply for any appreciated insight!