dwell time post analysis errors

[jc-brandeis]

The following errors appeared while trying to do the dwell time analysis using all default parameters with 2 exponential fitting. I am using version v1.1.17.post.dev4. I have not seen these errors with my other data previously. It could be data-specific.

Saved off-rate parameters in cosmos_dwelltime-koff-channel0.csv file
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
File ~/anaconda3/envs/tapqir-env/lib/python3.8/site-packages/tapqir/gui.py:1350, in dtCmd(b, layout, out)
   1348 with out:
   1349     logger.info("Dwell-time analysis ...")
-> 1350     dwelltime(
   1351         **layout.kwargs,
   1352         progress_bar=tqdm_notebook,
   1353     )
   1354     logger.info("Dwell-time analysis: Done")
   1355 out.clear_output(wait=True)

File ~/anaconda3/envs/tapqir-env/lib/python3.8/site-packages/tapqir/main.py:1280, in dwelltime(model, K, cuda, num_samples, num_iter, progress_bar)
   1276     torch.set_default_tensor_type(torch.FloatTensor)
   1278 fig, ax = plt.subplots()
   1279 ax.hist(
-> 1280     bound_dwell_times(
   1281         count_intervals(
   1282             model.params["z_map"][: model.data.N][
   1283                 None, model.data.mask[: model.data.N], :, c
   1284             ]
   1285         )
   1286     )[0],
   1287     bins=100,
   1288     density=True,
   1289 )
   1290 t = torch.arange(bound_dt.max())
   1291 y = 0

File ~/anaconda3/envs/tapqir-env/lib/python3.8/site-packages/tapqir/utils/imscroll.py:120, in bound_dwell_times(intervals)
    118 max_count = value_counts.max()
    119 n_values = len(value_counts)
--> 120 data = np.zeros((n_values, max_count), dtype=np.float32)
    121 for i in range(n_values):
    122     mask = result["posterior_sample"] == i

TypeError: 'float' object cannot be interpreted as an integer

[ordabayevy]

That looks like a strange error to me (max_count and n_values should always be integers). Does it happen with other dataset when you fit with 2 exponentials?

[jc-brandeis]

Hi Yerdos, There is some wrong with the channel 0 data analysis. There are no (or only a few) events in channel 0. Channel 0 data is very noisy because of high concentration of dye-labeled protein used in the experiment. That might have caused the error. I will look into the analysis more when the GPUs free up. Johnson

…

On Fri, Apr 14, 2023 at 12:29 PM Yerdos Ordabayev ***@***.***> wrote: That looks like a strange error to me (max_count and n_values should always be integers). Does it happen with other dataset when you fit with 2 exponentials? — Reply to this email directly, view it on GitHub <#422 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A2NW2SIHBTAGMH4VSVGICW3XBF3P3ANCNFSM6AAAAAAW6MH77Y> . You are receiving this because you authored the thread.Message ID: ***@***.***>