Re: Scipy

[Jose Guzman]

Hi Tobias!

Nice to hear from you. I am also very happy that you like Stimfit. I load custom files exactly as you do (see here https://neurodroid.github.io/stimfit/howto/ascii.html), but if the format is relevant - and I am somehow sure it is - you could contact Alois Schloegl from IST (alois.s...@ist.ac.at). He developed BioSig to import/export different formats for bioelectrical signals and we use it in Stimfit. If he implements the loading of your file in BioSig, you will immediately have it in Stimfit. 

Hope it works for you, let me know if I can be  of any help.

Cheers

Jose

On Tue, 5 Mar 2019 at 11:27, Tobias Rose <tro...@gmail.com> wrote:

Hi Jose, hi Christop,

first of all: thanks a lot for Stimfit - I played around with it and it is really helpful for our current mini analysis (Stimfit current MacPorts version).

One thing, though: The data we are analyzing is in a pre-HDF5 Matlab format (Janelia Ephus). Currently, I simply export data to txt and import it in Stimfit with a custom script. This is a step I'd like to skip, though. 

Is there an easy way of post-hoc including non-supported packages like Scipy in the MacPorts installation (python newb here)? I would need SciPy's loadmat function since pytables, which is included by default, is of not help much here.

Thanks,

Tobias

--

---
You received this message because you are subscribed to the Google Groups "stimfit" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stimfit+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Tobias Rose]

HI Jose,

Thanks for your response - I managed to get Scipy up and running (and deleted my original post in shame because it was so obvious ;) ). Like this I can directly load the Matlab structure and parse it for sampling rates etc. Works like a charm. Your software is really great for our mini analysis. I played around a bit with Paninski’s Bayesian event inference method (https://www.ncbi.nlm.nih.gov/pubmed/27208694) - but getting the parameters sufficiently tuned turned out to be a nightmare… Your workflow for template-based deconvolution with manual curation is really nice and straightforward - saves me a lot of programming work ;)

Best,

T

[Jose Guzman]

Hi Tobias,

thank you for your kind words. You must thank Christoph, who implemented the whole workflow based on the algorithm described in Pernia et al., 2016 [1] together with  Alois.  Before Christoph's implementation, the whole procedure was a complete nightmare, and it was difficulty reproducible by other labs.

Cheers!

Jose

[1]  https://www.ncbi.nlm.nih.gov/pubmed/23062335