search

pgleeson / neurotune

BSD 3-Clause "New" or "Revised" License
1 stars 0 forks source link

Analyzing voltage-clamp traces #5

Open VahidGh opened 9 years ago

VahidGh commented 9 years ago

@pgleeson, I'm working on a same approach taken by @vellamike for analyzing traces and spikes in pyelectro. It came to my mind that it would be very helpful to do some pre-processing on digitized data and estimate initial values for activation/inactivation tau, slope factor etc from a voltage-clamp trace, and then passing these values to the optimizer in order to decrease searching space. I could do this for activation parameters (as it is easier due to starting with stimulation), but it is a bit difficult when comes to inactivation. The problem is with finding the "right point", inactivation gate is closed, and the current going to steady state in a typical curve. May I have your idea, please? And also is it useful to put such a functionality in pyelectro/neurotune?

pgleeson commented 9 years ago

If it's something that's quite channel plot specific, you could indeed create a VClampAnalysis equivalent to IClampAnalysis, with methods specifically for analysing Vclamp data. If it's to go in pyelectro, it should be quite generic. However, some of the other logic you describe, inferring some max/min values from an analysis, would probably be very specific to the particular data set being used, and might be better in a repository for analysing one or more data sets.

Best to write the code first outside of pyelectro/neurotune and then see if some or any of it can be made into generic functions for wider reuse.

VahidGh commented 9 years ago

@pgleeson, I tried using analysis, min-max, etc functions from pyelectro, but returns no value for my VClamp trace (although traces are similar!). Will work on what you suggested. Thanks.

pgleeson commented 9 years ago

max_min and other analysis methods in pyelectro assume they are analysing data that contain spikes within the range of normal physiological data. A whole new set of analysis routines would have to be written for data of other types...