quality scores and the use of pandaseq to join paired-end Illumina reads in QIIME?

[Seaver Wang]

Hi all,

I've been trying to use pandaseq in conjunction with QIIME to assemble paired-end Illumina reads prior to performing quality filtering and demultiplexing.

What I'm essentially trying to do is use pandaseq in place of join_paired_ends.py as my assembly method (in order to retain more reads), then move on to demultiplexing using split_libraries. I've been having some trouble with this, namely with figuring out how to make my pandaseq output usable with QIIME. 

From what I can understand, pandaseq changes the meaning of the quality scores and therefore the output can't be used to perform quality filtering post-assembly.

As of now, I'm a little confused about how to design my workflow to work with pandaseq. I should join my reads prior to using split_libraries or split_libraries_fastq for quality filtering and demultiplexing, yet if I join my reads using pandaseq, the quality scores are no longer usable for quality filtering. Am I interpreting this correctly? Is there potentially another approach to what I'm trying to do?

I am quite new to QIIME, so please forgive me if I'm missing something obvious!

Best,

Seaver

[Colin Brislawn]

Hello Seaver,

Thanks for getting in touch with us. There are multiple reasonable ways of performing each step in the qiime pipeline. While I can't comment well on pandaseq (never used it), I've got an approach to suggest.

"Is there potentially another approach to what I'm trying to do?"

How about this: 

Join your reads with pandaseq.

Quality filter your reads with pandaseq. 

Use the qiime script add_qiime_labels.py to make a final seqs.fna file.  http://qiime.org/scripts/add_qiime_labels.html

In this method, you will state in the methods section that you use pandaseq for all preprocessing and quality filtering, then proceeded to OTU picking with qiime. I do something similar with the quality filtering tools of vsearch, and it's totally acceptable. 

Let me know if this method will work well for you,

Colin

[Seaver Wang]

Thanks for your answer Colin! I'm attempting this in conjunction with another potential solution at the moment, and I'll hopefully update shortly on how well this worked!