How to use QIIME process MiSeq 16s data?

[zhaopeihua]

Hi:

I’m totally new to this area, I've got some MiSeq 16S data. Is there tutorial like 454 Overview Tutorial("http://qiime.org/tutorials/tutorial.html") on QIIME website?

thanks in advance

[GenoMax]

http://qiime.org/1.6.0/tutorials/illumina_overview_tutorial.html

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3400413/

[zhaopeihua]

http://qiime.org/1.6.0/tutorials/ill..._tutorial.html

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3400413/

thanks for your reply

I hava checked this tutorial, it used for single-end,but I want to process pair-end data.
Do you know other tutorial or approach to do this? thank again.

[fanyucai1]

qiime could not cope with pair-end datas,but if you use miseq platform ,maybe there is a overlap between pair-end data. so you can assembly the data ,then use the qiime.

Hi:

I’m totally new to this area, I've got some MiSeq 16S data. Is there tutorial like 454 Overview Tutorial("http://qiime.org/tutorials/tutorial.html") on QIIME website?

thanks in advance

[csquared]

Very easy to use PANDASeq to assemble the paired reads and then use the assembled file in QIIME.

If you can use MacQIIME, I have a simple python script and some BASH scripts that make processing hundreds of samples very simple using default settings. Perfect for getting to know the software tools and then tweaking as you get more familiar with the tools and settings.

[Vesperholly]

Hi there,
We are looking to combine several runs of 16S Miseq data into one large data set for analysis - 400 to 500 samples all told. If you have any ideas about streamlining Qiime analysis for a set this size, or if there are any specific trouble spots we should keep an eye out for, I'd love to hear about them!

[rhinoceros]

Hi there,
We are looking to combine several runs of 16S Miseq data into one large data set for analysis - 400 to 500 samples all told. If you have any ideas about streamlining Qiime analysis for a set this size, or if there are any specific trouble spots we should keep an eye out for, I'd love to hear about them!

I've made a pipeline for just that, but it's work related and I can't post it. Anyway, it's a rather simple bash script that anyone could write. Everything starts from a map file from where relevant information is parsed and passed on to mothur for denoising. Then some header editing so every sequence surely has a unique ID (which is nice if you want to combine samples later on). Then back to QIIME for open reference otu picking. Nothing complicated. Almost everything QIIME related works just fine with default settings, however, in my experience more memory should be allocated to RDP-classifier, or you'll risk it hanging.

edit. My pipeline is for 454-data. Anyway, I'm going to adapt it for MiSeq data someday soon. Don't think much needs to be changed, just the preprocessing steps..

[bstamps]

You know there is a pre-existing way to have multiple runs of 454, Illumina...whatever... within QIIME, right? Just add -n some integer to the split_libraries.py or split_libraries_fastq.py that exceeds the number of reads in the first library. For example, for three Illumina runs with overlapping barcodes, and 10 million reads per library

split_libraries_fastq.py -i run1_reads.fastq -b run1_barcodes.fastq -m run1mapping.txt -o Split1Out/

split_libraries_fastq.py -i run2_reads.fastq -b run2_barcodes.fastq -m run2mapping.txt -n 10000001 -o Split2Out/

split_libraries_fastq.py -i run3_reads.fastq -b run3_barcodes.fastq -m run3mapping.txt -n 20000002 -o Split3Out/

Note that "-n" increments as you add libraries to some arbitrary number that is larger than the total possible number of reads in the previous library. This is done to number each sequence in the output seqs.fna so that there are no overlapping read names in each file.

Afterwards you would take the output from each SplitOut folder and concatenate them into a single seqs.fna

cat Split1Out/seqs.fna Split2Out/seqs.fna Split3Out/seqs.fna > seqs.fna

After this, make a new mapping file with all your samples in it.

And then run the remainder of your QIIME workflow as you want. You'll get errors about duplicate barcodes, but post split_libraries this is irrelevant.

See http://qiime.org/scripts/split_libraries_fastq.html for more guidance for this command, or here http://qiime.org/tutorials/denoising_454_data.html and here http://qiime.org/scripts/split_libraries.html if you have conventional or 454 libraries.

I've made a pipeline for just that, but it's work related and I can't post it. Anyway, it's a rather simple bash script that anyone could write. Everything starts from a map file from where relevant information is parsed and passed on to mothur for denoising. Then some header editing so every sequence surely has a unique ID (which is nice if you want to combine samples later on). Then back to QIIME for open reference otu picking. Nothing complicated. Almost everything QIIME related works just fine with default settings, however, in my experience more memory should be allocated to RDP-classifier, or you'll risk it hanging.

edit. My pipeline is for 454-data. Anyway, I'm going to adapt it for MiSeq data someday soon. Don't think much needs to be changed, just the preprocessing steps..

[Brajbio]

Check out this tutorial for Pre-processing Paired end Illumina data for QIIME

https://sites.google.com/a/brown.edu/bioinformatics-in-biomed/qiime-tutorial-3