tabulate-seqs reports the number of sequences in the input, not the number of rows in the table

[gregcaporaso]

If tabulate-seqs is called with --p-merge-method intersect or --p-merge-method union, the Sequence Count reported at the top left (see the star in the screenshot below) is the number of sequences in the input data Artifact. That may be different than the number of sequences reported on in the table (i.e., the number of table rows), which can be lower (if --p-merge-method intersect) or higher (if --p-merge-method union). The sequence length stats presented are, however, computed on the total Sequence count that is reported.

See these lines:

q2-feature-table/q2_feature_table/_summarize/_visualizer.py

Line 51 in a4d0e2e

seq_lengths.append(seq_len)

q2-feature-table/q2_feature_table/_summarize/_visualizer.py

Line 75 in a4d0e2e

seq_len_stats = _compute_descriptive_stats(seq_lengths)

and notice that seq_lengths isn't filtered in the --p-merge-method intersect case.

In the --p-merge-method intersect it would probably make the most sense to have the Sequence Count and length stats reflect the sequences that are reported on in the table.

In the --p-merge-method union case on the other hand, it's possible that we won't have sequences for some of the table rows, in which case the sequence length stats would reflect only what's present in the input data Artifact, so it probably makes sense to have Sequence Count reflect what's in the input data.

In both cases, we should be making it more clear specifically what Sequence Count is a count of.