different pore shapes from hole and hole2 on the same protein

[Rulong. Ma]

Hi MDnalysis developers,

I got two different pore shapes from hole and hole2 on the same protein. the issue discription is in the attecment.  if you want to know more, please let me know. thanks

[Oliver Beckstein]

Hi Rulong,

On Mar 15, 2021, at 8:55 AM, Rulong. Ma <marulo...@gmail.com> wrote:

I got two different pore shapes from hole and hole2 on the same protein. the issue discription is in the attecment.  if you want to know more, please let me know. thanks

HOLE is not a deterministic algorithm. It uses random numbers. If you want reproducible runs you can set raseed to a number and then you should get the same answer between two different runs. That said, the procedure is also sensitive to your initial search start (CPOINT) and direction (CVECT). Although the defaults in MDAnalysis work well most of the time, you know your protein better than anyone else so you might have to tweak these parameters. 

I suggest you read the original HOLE papers (see citations in https://userguide.mdanalysis.org/stable/examples/analysis/polymers_and_membranes/hole.html) and HOLE docs at http://www.holeprogram.org/doc/index.html and then you have a better idea how to drive HOLE. MDAnalysis just provides a wrapper around HOLE.

Oliver

--

Oliver Beckstein (he/his/him)

email: orbe...@mdanalysis.org
twitter: @orbeckst

MDAnalysis – a NumFOCUS fiscally sponsored project

https://www.mdanalysis.org/