Does the Mixed samples effects the IsoSeq Analysis?

[huan]

Hi, pacbio
Recently we got the mixed wheat samples with different treatment and used the SMRT(Pacific Biosciences) sequencing system PacBio RS II to generate cDNA sequences for the mixed sample. We have used P_IsoSeq to do the analysis and finally we got about 60,000 consensus sequence. Then I used the tofu collapse_isoforms_by_sam.py (--dun-merge-5-shorter) to remove redundant transcripts. But I find many groups have more than 20 sequence(one of the group have even more than about 8,100 sequence) which results in the nonredundant transcript number down to 9,000. At the same time, we also sequecing a wheat sample using Pacbio and do the same analysis. But we get about 40,000 nonredundant transcript from 50,000 consensus sequence.
So I am confused is there any problem with my mixed sample experiment?

Thanks a lot for any advice.

[bowhan]

I have not seen a wheat sample before but I don't think mixing samples in general has an impact on this.

Is the mixed sample barcoded?
If they are, can you split them and run them individually?

How is the quality of the RNA (check coverage using picard CollectRnaSeqMetrics for example)?
If the first sample is more severely degradated than the second one, there are more artificial products.

Can you show a screenshot of the genome alignment of the gene with 8100 sequences in both libraries?