I didn't find an existing issue on that, so decided to write an issue of my own.
So, I would like to use suite2p to detect cell bodies in samples. But in those samples, instead of GCaMP6 we use Fluo-4. As I understand, the main difference here will be the decay time. The decay time of Fluo-4 fluorescence is short -- several hundred milliseconds to a few seconds, while for GCaMP6s it's seconds to tens of seconds. Maybe GCaMP6's sensitivity is also somewhat lower.
What I would like to ask is some advice on which parameters I should try adjusting to make suite2p work for other markers (not GCaMP). As of now, it doesn't seem to correct cell bodies at all correctly, so 100% of what it identifies, I have to manually put into "not cells" category, and then the classifier can't learn anything. Specifically, real cell bodies are bigger than it detects, i.e. it only marks for example a little border of the whole cell. It seems to point to the soma's size issue, but is it also possible to adjust decay time and is there something else needed?
I hope it's okay to put up such a question, it may be a little abstract or naive, but I would like to know your opinion on what parameters I need to adjust to make it work, and is it possible at all to make suite2p work with other markers.
I didn't find an existing issue on that, so decided to write an issue of my own. So, I would like to use suite2p to detect cell bodies in samples. But in those samples, instead of GCaMP6 we use Fluo-4. As I understand, the main difference here will be the decay time. The decay time of Fluo-4 fluorescence is short -- several hundred milliseconds to a few seconds, while for GCaMP6s it's seconds to tens of seconds. Maybe GCaMP6's sensitivity is also somewhat lower.
What I would like to ask is some advice on which parameters I should try adjusting to make suite2p work for other markers (not GCaMP). As of now, it doesn't seem to correct cell bodies at all correctly, so 100% of what it identifies, I have to manually put into "not cells" category, and then the classifier can't learn anything. Specifically, real cell bodies are bigger than it detects, i.e. it only marks for example a little border of the whole cell. It seems to point to the soma's size issue, but is it also possible to adjust decay time and is there something else needed?
I hope it's okay to put up such a question, it may be a little abstract or naive, but I would like to know your opinion on what parameters I need to adjust to make it work, and is it possible at all to make suite2p work with other markers.