16S Barcoding long and short sequences return different BLAST results

Question

[robert.murphy]

When performing 16S amplicon sequencing what should one do if the long and short sequences return different BLAST values both with high bit score, low E value, high % pairwise identity and 100% coverage.

My instinct would be to go with the result of the long sequences as they carry more information including the nested information of the short sequences when both have 100% coverage?

Short

Long

[colindaven]

I'd trust the long sequences more. But why do you have short and long sequences anyway ? Full length, or just limited to regions of 16S ? Are the full length from long nanopore or pacbio reads ?

[robert.murphy]

The short sequences are just regions where there is overlap from the forward and reverse read. Long reads are the total combination of both forward and reverse.