Pancreas counts are not integers

[rcannood]

import openproblems
pan = openproblems.data.pancreas.load_pancreas()
print(pan.layers['counts'])

(0, 16)       1.0019583
(0, 20)       1.0019583
(0, 25)       1.0019583
(0, 30)       2.0078535
(0, 40)       1.0019583
(0, 66)       4.0315795
:     :
(16381, 18633)        93.4915
(16381, 18641)        5.5119
(16381, 18660)        1.2677
(16381, 18661)        4.6227
(16381, 18665)        0.1375
(16381, 18675)        7.6811
(16381, 18681)        32.8928
(16381, 18684)        0.6472
(16381, 18687)        113.5626
(16381, 18698)        1.236
(16381, 18705)        0.1846
(16381, 18711)        3.3464
(16381, 18728)        0.2025

The .layers['counts'] matrix should be whole numbers instead of real numbers.

If there were scripts available for these datasets (related to #679) I'd be able to see where things went wrong.

[LuckyMD]

Look at the batch assignment of the non-integeger values. They may be from non-UMI technologies.

[scottgigante-immunai]

These are indeed the non-UMI technologies. This is expected behaviour. I agree, however, that was should have generation scripts.