Assertion error while using the PlaneCA1PC

[Gilles Vanwalleghem]

Hello,

First thanks for the package. I am trying to use it to analyze our nuclear targeted GCaMP (May the nuclear targeting be the issue ?). So I am trying to apply the tutorial to our dataset :

segmentation_approach = sima.segment.PlaneCA1PC(

    num_pcs=30,

    max_dist=(4, 4),

    spatial_decay=(2, 2),

    cut_max_pen=0.10,

    cut_min_size=50,

    cut_max_size=150,

    x_diameter=6,

    y_diameter=6,

    circularity_threhold=.5,

    min_roi_size=20,

    min_cut_size=40

)

dataset.segment(segmentation_approach, 'auto_ROIs')

And I end up with an assertion error :

Either on the local computer 

\Anaconda\lib\site-packages\sima\segment\normcut.pyc in calculate(self, dataset)

    405                             (y + dy < shape[0]):
    406                         w = self._weight(r0, r1)
--> 407                         assert np.isfinite(w)
    408                         a = x + y * shape[1]
    409                         b = a + dx + dy * shape[1]

AssertionError: 

or on a virtual machine :

/home/ubuntu/anaconda/lib/python2.7/site-packages/sima/segment/normcut.pyc in calculate(self, dataset)
    405                             (y + dy < shape[0]):
    406                         w = self._weight(r0, r1)
--> 407                         assert np.isfinite(w)
    408                         a = x + y * shape[1]
    409                         b = a + dx + dy * shape[1]

AssertionError: 

Do you have any advice on what to change ? I am trying to run other segmentation approaches and so far it looks like they are proceeding.

Thank you

[Patrick Kaifosh]

It's hard to tell you much just based on the error message. You could
try using pdb and see if you can figure out any clues (e.g. find out
what caused w to become non-finite)? I don't think the nuclear targeting
would be the issue there, but I wouldn't recommend the PlaneCA1PC for
nuclear targeted-GCaMP in general because it relies on an assumption
that the GCaMP signal is darker in the nucleus and brighter in the
cytoplasm -- definitely not true in your case. With nuclear GCaMP, I
imagine the segmentation problem should be a fair bit easier. However,
we haven't written any code for that case. Maybe you will?

Gilles Vanwalleghem wrote:
> Hello,
>
> First thanks for the package. I am trying to use it to analyze our
> nuclear targeted GCaMP (May the nuclear targeting be the issue ?). So
> I am trying to apply the tutorial to our dataset :
>
> segmentation_approach = sima.segment.PlaneCA1PC(
> num_pcs=30,
> max_dist=(4, 4),
> spatial_decay=(2, 2),
> cut_max_pen=0.10,
> cut_min_size=50,
> cut_max_size=150,
> x_diameter=6,
> y_diameter=6,
> circularity_threhold=.5,
> min_roi_size=20,
> min_cut_size=40
> )
> dataset.segment(segmentation_approach, 'auto_ROIs')
>
> And I end up with an assertion error :
> Either on the local computer
>

> \Anaconda\lib\site-packages\sima\segment\normcut.pycin calculate(self,

> dataset)
> 405 (y + dy < shape[0]):
> 406 w = self._weight(r0, r1)
> --> 407 assert np.isfinite(w)
> 408 a = x + y * shape[1]
> 409 b = a + dx + dy * shape[1]
>
> AssertionError:
>
> or on a virtual machine :
>
> /home/ubuntu/anaconda/lib/python2.7/site-packages/sima/segment/normcut.pyc in calculate(self, dataset)
> 405 (y + dy < shape[0]):
> 406 w = self._weight(r0, r1)
> --> 407 assert np.isfinite(w)
> 408 a = x + y * shape[1]
> 409 b = a + dx + dy * shape[1]
>
> AssertionError:
>
> Do you have any advice on what to change ? I am trying to run other segmentation approaches and so far it looks like they are proceeding.
>
> Thank you

> --
> You received this message because you are subscribed to the Google
> Groups "sima-users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to sima-users+...@googlegroups.com
> <mailto:sima-users+...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sima-users/9c6e9ef1-f8f3-494a-b04a-8dfe1dcae103%40googlegroups.com
> <https://groups.google.com/d/msgid/sima-users/9c6e9ef1-f8f3-494a-b04a-8dfe1dcae103%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.