Sunil - i'm not a frequent QIIME user, but I don't see any problems here. Merging paired end reads is done for mates, i.e. for each read in the forward and reverse FASTQ file there exist exactly one paring as indicated by the read header (ID, position, forward/reverse flag, ...). For that, it does not matter whether the first paring of your two FASTQ files are originally forward-reverse or reverse-forward mates. You will end up with one merged file where your reads are either in forward or reverse direction.

Then you can follow the usual workflow and start splitting the libraries based on the multiplex identifier. You might need to run the corresponding split_library script twice and provided each time a different combination of MID and forward and reverse PCR primers. So you will end up with two libraries for each MID, one beginning with the forward and one with the reverse primer. Then you can reverse transcribe the latter one to get them in the right orientation and merge them with the former one - viola (theoretically, i've never done this, but this approach should do it).

And of course, merging paired-end reads always depends on the insert size of your fragments, so you should check whether merging the mates is reasonable for you.

Thanks Gabrielle and Sebastian

Yes persent version pf both QIIME and MOTHUR cannot handle this problem, i asked in the forum they said it will be implemented in next version very soon.

Gabrielle your suggestion look interesting i never thought about using only one read, i am also working on ITS2 data amplified by ITS7a and ITS4 primer combination and expected product size in 375 bp and my forward read length is 300bp.

Sebastian, i am trying to do something similar what you have suggested, but i was trying to follow the mothur, here there is no proper option for making quality filtering if i have paired end miseq data. and in qiime its complicated issue of extract_barcode.py , as i didn get any from machine.

Regards

Sunil

You could attempt to use the RDP pipeline initial process, which can assemble paired end reads (pyro.cme.msu.edu) under pipeline initial process. I use the PANDAseq assembler via a high performance computing center where the python script is hosted, so I am not sure if it will work with your files, but it is worth a try.

Thanks all for your suggestion

Finally i managed to do joining in qiime and quality control in galaxy and then demultiplexing in qiime with F primer first and then reverse primer first and than reverse complementing to reverse output and join it with forward primer output.

It was bit complicated.....

Hi Sunil, I played quite a lot with joining fastq Illumina reads. Here is suggestion from Qiime team. It worked for my dataset. Approx half of the reads joined

Look here:

https://groups.google.com/forum/#!msg/qiime-forum/QTfDdB_fcJE/i20gpt7lWn0J

There is also other programs/scripts like PandaSeq or PEAR wich should do the trick for you as suggested earlier.

Other think. I tried also makecontigs in mothur. But mothur joined ALL reads, so I was suspitios about the ouput...