search

hiclib / pastis

Poisson-based algorithm for stable inference of DNA Structure
http://hiclib.github.io/pastis/
Other
34 stars 15 forks source link

Does Pastis have a parameter for specifying nucleus size? #50

Open danextreme opened 4 years ago

danextreme commented 4 years ago

Thanks for this code. i really love it ! my question is, Does Pastis have a parameter for specifying nucleus size? There seems to be a parameter that specifies nucleus size in the config.py file. however in the algorithms.py file, it appears it is not utilized by any of the algorithms. I also tried specifying the nucleus size in the config.ini file, but it results in an error ("TypeError: cannot create 'NoneType' instances"). I am wondering if it is possible to include a constraint that accounts for the nucleus of the cell when solving the 3D structures in Pastis, and specifically i want to know how to specify this parameter (with example, please). Thanks in advance.

NelleV commented 4 years ago

Hello, If the nucleus size is available, the only element it would do is rescale the structure after the inference so that the resulting structure is at the correct scale. In practice, the inference is entirely identical at any scale. But the problem you are facing with the option in the config.ini should not happen anyways… I'll check this out.

For now, I would recommend doing the inference without the nucleus size, and rescaling the structure afterwards.

danextreme commented 4 years ago

Thanks for the quick response.

How does the nucleus_size parameter differ from the beta parameter for scaling the 3D structure?

Also what is the unit for the nucleus_size parameter? nanometers? micrometers?

Thanks in advance

NelleV commented 4 years ago

Hi, Sorry for the slow reply. As mentioned before, I think you should ignore this parameter, and rescale the resulting structure to match your prior knowledge of the nucleus_size. Thanks, Nelle