virtual4c assigning nan to point of view

[claudiarabu]

Hi!

I was using the virtual4c function of cooltools and noticed that the point of view is assigned as nan, but this might lead to some confusion as also null (= not happening) interactions are assigned as nan. Also, for plotting I substituted every nan with 0 values, but the resulting graph at around the pov bins was of course a bit misleading.
What I did was changing the cooltools/api/virtual4c.py to assign value 1 to the pov bins, but I guess another option could be to assign 0 to null interactions and nan to the pov bins.

Besides that, cool tools indeed! :)

[Phlya]

Mh can you clarify in some more detail why setting bins within the viewpoint to Nan is misleading? I think that should at least be the default behavior...

[claudiarabu]

Yes, I agree that assigning no interactions to 0 and nan to the pov bins would be the best option!

P.s. I still have nan in case of no interactions, but maybe that was not intended and it caused the discussion!

[Phlya]

Are you sure it's for the case of no interactions and not for the case of masked bins?
If so, it's a bug...

[claudiarabu]

Well, yes, unfortunately I am sure and I guess it's a bug.

[Phlya]

Ok, thank you, I'll look into it next week then! On holidays until then.

[claudiarabu]

Hi Phyla,
Long story short: it's not a bug. I needed to manipulate my sample before converting it to mcool to adjust the interaction values of some regions with a different ploidy and this caused many bins to be considered as masked in the virtual4c, but if I don't manipulate the sample I do get 0s and nan as separate values. Sorry for the confusion and enjoy your holidays!