How is track height determined?

[KewinOgink]

Hello, I noticed that I have a region with several SNP tracks, where I know the amount of SNPs per track (written here for 2.3-2.4Mb and I use the same bin width of 1kb for all tracks. However, the number/sum of variants is not matching the visualization:
.

I then made a dummy file to investigate using bin width 4 and 3 tracks with different densities

%%writefile ../tmp/plotsr/fake_1_per_10.bed
chrA01	0	1
chrA01	10	11
chrA01	20	21
chrA01	30	31
chrA01	40	41
chrA01	50	51

%%writefile ../tmp/plotsr/fake_2_per_10.bed
chrA01	1	2
chrA01	2	3
chrA01	10	11
chrA01	11	12
chrA01	20	21
chrA01	21	22
chrA01	30	31
chrA01	31	32
chrA01	40	41
chrA01	41	42
chrA01	50	51
chrA01	51	52

%%writefile ../tmp/plotsr/fake_10_per_10.bed
chrA01	0	10
chrA01	21	30
chrA01	41	50

%%writefile ../tmp/plotsr/fake.out
chrA01	0	1000	-	-	chrA01	0	1000	SYN1	-	SYN	-

%%writefile ../tmp/plotsr/fake_w4.track
#file	name	tags
../tmp/plotsr/fake_1_per_10.bed	fake_1_per_10	ft:bed;bw:4;nc:black;ns:8;nf:DejaVu Sans;lc:sienna;lw:0.25;bc:lightblue;ba:0.5;tt:f;ti:1
../tmp/plotsr/fake_2_per_10.bed	fake_2_per_10	ft:bed;bw:4;nc:black;ns:8;nf:DejaVu Sans;lc:sienna;lw:0.25;bc:lightblue;ba:0.5;tt:f;ti:2
../tmp/plotsr/fake_10_per_10.bed	fake_10_per_10	ft:bed;bw:4;nc:black;ns:8;nf:DejaVu Sans;lc:sienna;lw:0.25;bc:lightblue;ba:0.5;tt:f;ti:3

and ran plotsr like this

min_sv_size=0
chrom=chrA01
track_window_size=4
ml plotsr

plotsr \
    --sr ${wd}/fake.out \
    --genomes ${wd}/../${sample}.genomes.txt \
    --tracks ${wd}/fake_w${track_window_size}.track \
    --reg ref:$chrom:1-60 \
    -S 0.4 -o R7 -W 10 -H 5 -f 8 -s ${min_sv_size} \
    -o $wd/plot_test_${chrom}:1-60_s${min_sv_size}_tracks_w${track_window_size}_h.png

result:

Given the snp density of track 2 is 2x that of track 1, why is the pattern the same? And why does track 3 not show a full brown block, as no matter where the bin starts, it 10 bp spans the bin 2.5x.

I then ran it with bin widht 1 and then I do get this:

This result that looks correct. However, this is not feasible for the entire genome. Any suggestions? I was not sure how/if this relates to #34

Many thanks!

[mnshgl0110]

For each track, the height is normalised by the maximum value in that track, i.e., the y-axis for the tracks are on different scales. Fake1 and fake2 have same apparent pattern for BW=4 because the relative distribution along the chromosome is same.

Regarding fake3, consider the row ChrA01:21-30, the corresponding bins of interest would be [20, 21, 22, 23] , [24, 25,26, 27], and [28, 29, 30, 31] and for them the bin value (for the given range in BED file) would be 0.75, 1, and 0.5 respectively (note: in BED file, the start position is included but the end position is not). Therefore, the tracks for ChrA01:21-30 and ChrA01:41-50 are correct. The track for ChrA01:0-10 is wrong and corresponds to ChrA01:1-10. Could you please recheck that the plot that you shared corresponds to the given tracks.

[KewinOgink]

Thanks for the explanation. It makes sense. So if in fake1 1/4 is the highest relative value, this will be ymax, whereas if for fake2 it is 2/4 this will be ymax.

For me it is relevant to have the tracks sharing the same y axis. Therefore, one working (though not ideal) solution is to add a fake full window. I tested it to be both outside (100-110bp) and inside (51-60bp) the plotting range (0-60bp), and found that only if I have the fake full window within the plotting range it works, like this:

%%writefile ../tmp/plotsr/fake_1_per_10.bed
chrA01	1	2
chrA01	11	12
chrA01	21	22
chrA01	31	32
chrA01	41	42
chrA01	51	60  

%%writefile ../tmp/plotsr/fake_2_per_10.bed
chrA01	1	2
chrA01	2	3
chrA01	11	12
chrA01	12	13
chrA01	21	22
chrA01	22	23
chrA01	31	32
chrA01	32	33
chrA01	41	42
chrA01	42	43
chrA01	51	60

%%writefile ../tmp/plotsr/fake_10_per_10.bed
chrA01	0	10
chrA01	20	30
chrA01	51	60

And indeed I think perhaps one coordinate was wrong as you pointed out.

[mnshgl0110]

I agree, that there is merit in having same yaxis across different tracks. Might be possible to add this feature in future updates.

[mnshgl0110]

The commit adds norm config option for base.cfg that allows to toggle track height normalisation.