Cell Number inconsistency

[michaelbornholdt]

sc-metadata file after subselecting training and 0.5 sampling rate:

BRD-A09533288-003-31-2    3294
BRD-K92968657-001-01-6    3294
BRD-K89375097-300-08-8    3294
BRD-K14704277-001-22-6    3294
BRD-K16542329-001-13-5    3294

Counting the location files in the input from BRD-A09533288-003-31-2.
= 10598
This is sub_index. After this, one has to 3/5 * 1/2 -> checks out actually.

For a different sample, the value was similar but not the same:
11437
SO there is something done in the sampling.

The Cell count values from LINCS repo tell a different story tho:
95382.0 for that sample.