Same problem with data from Mr. DNA.  I've lost a week of my life emailing the company and trying to get help, only to be repeatedly redirected to a hodge-podge set of "instructions" clearly cobbled together from the Illumina documentation.  We're using a different sequencing facility for all future runs, as this is the first time we've had this kind of problem.

Hi Ciara,

I would try to use QIIME to do these initial steps using the script join_paired_ends.py

see http://qiime.org/scripts/join_paired_ends.html

and below for a guide on full processing of illumina data-

http://qiime.org/1.2.1/tutorials/processing_illumina_data.html

Hope that helps.

Hi Joe,

Thank you for your suggestion. I am now starting to worry that the R1 and R2 files contain a mix of forward and reverse reads which may explain why I can't get them to work. I have QIIME installed so I will try this first step there now and see does this work.

The information from Mr. DNA 4. To process the r1 and r2 files for amplicons 250bp (subjective). And process as usual.

c. The same basic process is used by MR DNA for amplicons too long to be joined effectively.

Remember if not joining the reads (which reorients the barcodes in the r2 files) both files can be used for finding indexes (barcodes) without reverse complementation. After binning the R1 file you can then bin the R2 file to identify all of the barcoded reads.. their pairs can be identified using the illumina definition line coding, reads can be joined with appropriate software if available.. MR DNA does now offer FREE joining of paired reads for amplicon sequencing as well as all of our comprehensive taxonomic analysis, advanced analysis and full report service. )

For assays that have amplicons >300bp and < 570bp joining is very efficient. For assays that are longer or shorter than this we will attempt joining. if joining is inefficient due to the size of the amplicon we will analyze unidirectional reads."

Best wishes,

Ciara

I dont use mothur because I have my own programs.

For assembling reads I tried several tools, I selected two.

Flash is very very fast (use mutliple cores).

Pear is not that fast but as good.

Depending on the run, one or the other has a better assembly yield.

Usually best is around 90% (but this will depend on your amplicon len)

The first step should be an estimate of you min, mean and max amplicon len using some reference database (16S or 18S ?).

Cheers

Hey, can you upload the command lines you used in your analysis? I looked into make.contigs (http://www.mothur.org/wiki/Make.contigs), it doesn't necessarily need a stability.files, you can try 

mothur > make.contigs(ffastq=R1.fastq, rfastq=R2.fastq)

also, i think it's the company's responsiblity to send you a merged reads file. Can you ask them for that?

Hi everyone, 

Thank you for the suggestions. Pat Schloss himself responded to my thread in the forum and emailed me some suggestions. He suggested only using the R1 read due to the quality of the R2 read. I did manage to make.contigs using the command (I was using that all along but it wouldn't identify samples) but the results downstream were terrible....I think I will go back and try CutAdapt and then try re-run. For the moment I'm going to proceed with the R1 read and see how the results turn out. 

I'm afraid QIIME isn't an option right now as I have to submit my thesis soon and to learn a new program from scratch is a bit daunting and I'm not sure I would have the time. I did attempt the command in QIIME after getting it installed but I could not get it to work. (Because of me not being very adept at command line operations). I also tried it on MacQIIME on my laptop..

Hi Emma,

It was a bit complicated and  I had to leave this analysis until after thesis submission.When I used just the R1 read the results were very strange with 200,000 unique sequences and the alignments wouldn't work. But the problem is that the R1 and R2 files have a mix of forward and reverse reads so I think that's what they will not work with the Mothur commands.  I had been speaking to Sunil Mundra (he's on research gate) 

He recommended;

"In this case i would suggest to try joining of your sequences in using fastq-join (join_paired_end.py in QIIME) and then quality control in galaxy.

Convert in 454 sequence format (this things required if you have variable length barcode as split_library.py can function with variable lenght, whereas split_libraries_fastq.py (designed for Miseq analysis) cannot handle variable length barcode)

Use demultiplexing script of 454 (split_libraries.py), using forward primer as LinkerPrimerSequence - output 1

Repeat demultiplexing script of 454 (split_libraries.py), using reverse primer as LinkerPrimerSequence - output2

output 3 - Reverse complement output 2

Join output 3 + output 1 (now things should be in same order)

Now afterwards you can follow normal 454 protocol for further analysis."

I did not manage to get this to work as one of the outputs was blank but I think as I had never used QIIME before that it was simple mistakes that I was making. Because in theory that suggestion makes sense. I'm hoping to start tackling the data again in the next few weeks. 

I'd be very interested to see how you get on with trying to analyse your data, perhaps we could help each other with some suggestions.

Ciara

Hello Clara,

I have the same problem with ITS... when I use fastq-join (join_paired_end.py in QIIME)  the number of reads was very low (18000) and R1 and R2 files have 190.000 reads...how I can solve it? Thanks

Hi Ciara,

You need to prepare a tab- delimited file containing sample name and the corresponding forward and reverse fastq file names.  You should upload that file in the same directory as your read files and rename it to have the extension as ".files".  Then run make.contigs(file=filename.files, processors=8)

Hi Ciara,

I was trying to use the make.contigs step in Mothur with raw data (R1 and R2) from illumina Miseq and I had similar problems than you. I also received two files in fast.q extension with pooled forwards and reverse reads of various samples from the sequencing company Mr. DNA. Our assay have amplicons >300bp and

Same problem with data from Mr. DNA.  I've lost a week of my life emailing the company and trying to get help, only to be repeatedly redirected to a hodge-podge set of "instructions" clearly cobbled together from the Illumina documentation.  We're using a different sequencing facility for all future runs, as this is the first time we've had this kind of problem.

Hi Dr. Rumpf,

I think Mr. DNA is fine if you don't intend to do any analysis with the raw files, for some people this is ok-in the end it had to be ok for my PhD. But it should be made clear that the format the reads are in is not R1 Forward and R2 Reverse, they definitely changed the read me file after I sent numerous emails-I wasted so much time trying to make sense of it! I believe Sunil (comments above) got them to work through QIIME but it is a pain!

We had sequencing done with Research Technology Support Facility Michigan State University, the cost was very very cheap and the sequence coverage was really good plus the customer service (we dealt with Dr. Kevin M. Carr). 

There should be a Rate My Sequencing Facility webpage!

Best wishes,

Ciara

Just go for the single read approach. That's a safe bet.

Btw, why do you say the downstream results were terrible?

Hi Xabier,

The issue is that forward and reverse reads are in each of the R1 and R2 file actually so to proceed to only use the forward reads you need to separate forward and reverse from both files. Which at the time of writing this question (two years ago), I did not realise the issue as make.contigs never would have worked with an R1 and R2 file containing a mix of forward and reverse reads. 

Best wishes,

Ciara

Hi All, in the mean time of these thread, Is it possible to use Mothur for further analysis of diversities, etc-with files where both R1 & R2 joined, trimmed, and provided to me in a fasta-qual-mapping? I see from these files, the company went on defining OTUs and taxonomic classification from GreenGenes.... I grouped sequences from each samples together by Bioedit. what syntax may I use from this step to get into Mothur? 

I have had all sorts of problems with mothur...I ended up using Flash to make contigs because mothur has weird text requirements for the sequence headers and filenames. It was easier to use something I trusted rather than reformat all my data to work with mothur's code

also Qiime is much easier to pick up than Mothur, so I wouldn't worry about the time to learn the new software

I think before starting work on the real data of the company you should move through the qiime2 tutorial. It starts from single read sequences with forward and reverse read, barcodes and other essentials associated with it. I also got raw data from the company, I was very confused for two days, then I thought, I must go through the qiime2 protocol finally I did analysis myself.