Miseq reporter demultiplexing

[sunleaf]

Hi, All:

New to the forum here. I was running Miseq PE with 1 indexing (read1 14 cycles, index 6, read2 50 cycles), and the run finished without issue. However, during Miseq reporter analysis, only the 50 cycle reads were demultiplexed and fastq generated, while no 14 cycle reads fastq was generated.

This is what the sample sheet looks like:

Workflow,GenerateFASTQ
Application,FASTQ Only
Assay,TruSeq LT
Description,
Chemistry,Default

[Reads]
14
50

[Settings]

[Data]
Sample_ID,Sample_Name,Sample_Plate,Sample_Well,I7_Index_ID,index,Sample_Project,Description
sample1,1,sample plate,A01,A006,GCCAAT
sample2,2,sample plate,A02,A012,CTTGTA

Has anyone experienced similar issue? Do I need to use CASAVA to do my own demultiplexing? Thanks a lot!

[kcchan]

It may have something to do with RTA not getting enough data to perform basecalling. The first 12 bases are used to calculate the color matrix and phasing/prephasing values, and 25 are needed for quality scoring.

Just curious, may I ask why this run is designed to be in this asymmetric manner?

[sunleaf]

Thanks for your answer. That's what I suspected too. We are developing our own assay and 14 cycle is all we needed for that end. I'll contact Illumina to see if I can salvage the data and run 25 cycles in the future.

[kcchan]

You can check the error logs in MSR and see what's going on. I would guess that BCL files were never generated from read 1 and all you had were CIF intensity files. You might be able to use offline basecaller to get the basecalls out of it, but I've never tried it in such a situation. Illumina tech support may be your best bet in this case. Good luck and let us know how things turn out!

[sunleaf]

Just talked with a very helpful Illumina tech support. Yes, we can use CASAVA to demultiplex the other end with --usebasemask option and designated 14 bases, and the sample sheet should be made for Hiseq instead of Miseq. That's just too much trouble to go through for our repeated runs. I think it would just be easier to run read1 for 25 cycles.

[kmcarr]

Just talked with a very helpful Illumina tech support. Yes, we can use CASAVA to demultiplex the other end with --usebasemask option and designated 14 bases, and the sample sheet should be made for Hiseq instead of Miseq. That's just too much trouble to go through for our repeated runs. I think it would just be easier to run read1 for 25 cycles.

If you are using a 50 cycle kit make sure that the total number of cycles doesn't exceed the capacity of the kit. Officially a 50 cycle kit has sufficient reagents for 76 cycles; we have safely done 78 cycles. If you increase read 1 to 25 cycles without changing the other settings you cycle total will be 81 (25|6|50).

[kcchan]

Good point kmcarr. It's never good to run out of reagents since it can suck air into the lines. I believe the scan mix is usually the first to run out.

It may be a good idea to re-think the library design a bit. You can squeeze the indices inline at the start of read 1 to lengthen the read, but unfortunately I don't think the MiSeq has an easy way to demultiplex inline barcodes, so you're stuck with CASAVA if you want to go that route.