Duplicate observation IDs error even no duplication was found

[mortonjt]

Issue was first reported here scikit-bio/scikit-bio#2121

@quliping let's continue the discussion here for future reference

[wasade]

Odd. Could be an evaluation order or in the errcheck. What happens if the table is loaded directly into biom without using pandas?

[quliping]

Odd. Could be an evaluation order or in the errcheck. What happens if the table is loaded directly into biom without using pandas?

Sory for the late reply. I tried Table.from_tsv(), and it worked. I'm not sure if I should use this command, but the table was loaded into biom successfully.

And I think I found the reason... I tried my orignal script, asv_table = Table(df.T.values, observation_ids, sample_ids) worked but asv_table = Table(df.values, observation_ids, sample_ids) got the 'Duplicate observation IDs' error. It seems like I made a mistake for the row and column contents. Each column should be a sample and each row should be a species, right? I got it wrong before (I inverted my table). I think the error message is confusing and perhaps should change to something like 'The number of columns is different from the number of given sample ids, please check if there is a mistake in the placement of samples and observations in given table.' Because there are indeed no duplicate ids.

[wasade]

A traditional feature table, dating from QIIME 1 days, has rows as observations and that is modeled in BIOM. That's correct though that the error message is wrong, as it seems this should trigger a shape or related error, and that would be valuable to fix. Thank you for reporting!