This is my personal opinion, but I have seen that Mothur produces more realistic data like for eg: OTU compared to QIIME, this has already been shown in the couple of papers by P.Schloss, and to answer Sunils question about demultiplexing sequences within mothur, this has been outlined here in the 454 SOP

http://www.mothur.org/wiki/454_SOP#OTU-based_analysis

the exact command is http://www.mothur.org/wiki/Trim.seqs, using an oligos file.

The logical steps would be make the contigs the follow a trim.seqs and carry on with the miseq SOP.

Hope that helps.

Bottom-line I rather use Mothur with a silva reference data base than qiime with greengenes.

I would first perform trimming based on Qscore. CLC Bio can do that. Qscore > 20, length of >36 bp.

Once the trimmed data are out, map them against the human genome and compare the SNPs / INDEL frequency.

Hi Christopher,

Han's suggestion is a good one. However, as far as I remember, CLC Bio is not a free software.

I assume that you want to trim your raw data (i.e., the read sequences you get from the machine, normally in a bam or fastq format), right? I like to trim not only according to the Q value but also according to the clustering of low quality calls. To do so, I use the program ConDeTri (http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0026314). In the supplement of this paper you have very nice figures explaining the whole trimming procedure. This program works quite well and is free.

Hope this helps.

Yes to trimming of the raw data. But more on what software / commands can generate some good figures + other ideas for analysing the data...

I should probably have been more clear that I am interesting the the microbial ecology, so my samples have been amplified using universal bacterial primers and ultimately I want to compare the difference between bacterial profiles between healthy and diseased patients and also compare longitudinal within patient samples to track community change in the run up to and post disease diagnosis

Oh i see. In that case, firstly, overlap your paired end reads with FLASH or Pandaseq without even trimmming.

then blast against 16S microbial

then view it on MEGAN. MEGAN is really great for generating publicatio quality figure.

Why not follow the SOP for analysis that Pat provides? I wouldn't trim anything before you assemble the reads- you could use Pat's make.contigs or go with Pandaseq or Flash, but I'd stick with the mothur analysis for simplicity (it will make your group files etc.). Visualization will depend on your results (helps to have a knowledge of R!)

I agree with Robert, I use spotfire and R to make graphs and plots there are some R scripts and packages like Vegan that you can use, spotfire is commercial but good for data drilling ie visualisation.

alternately there are scripts around the net for cytoscape visualisation if that interests you.

Pat has detailed a lot about the miseq SOP in the AEM paper in case you have missed it.

There are several option:

- FastQC gives you an optimal overview of the produced sequencing data.

- if you want to overlap the paired-ends you can try Flash;

- Then you can explore the microbial diversity using a similarity-based approach (blast, or maybe a more fast program like Bowtie2, BWA and so on, plus tool like MEGAN) or an approach based on sequences characteristics, such as RDP-classifier

Did you try MGRAST or the EBI metagenomics portals to assess the bacterial diversity? These are web applications designed for next gen data and should give you a taxonomic profile of your reads as far as I know.

Dear folks,

what about Solid seq reads if you are having seq in csfasta and qual instead fastq ??

Hi all

I have rather similar question to community.

I have paired end illumina data.

If i follow Pats SOP in MOTHUR, i need sample to be demultipled based on tags first. There is nothing mention how to demultiplex the reads in mothur (F and R) before you can follow the SOP.

I have only two file R1 and R2 from machine and i have many different sample tag in the raw data. How can i proceed further?

Suggestion will be helpful

Regards

Sunil

I am currently into QIIME and I am impressed how much analyses you can perform with it. You can also have the possibility to interact with R and BIOCONDUCTOR and several related packages that are the state of art for statistical analyses and high quality figures.

Hope that helps,

G.

This is my personal opinion, but I have seen that Mothur produces more realistic data like for eg: OTU compared to QIIME, this has already been shown in the couple of papers by P.Schloss, and to answer Sunils question about demultiplexing sequences within mothur, this has been outlined here in the 454 SOP

http://www.mothur.org/wiki/454_SOP#OTU-based_analysis

the exact command is http://www.mothur.org/wiki/Trim.seqs, using an oligos file.

The logical steps would be make the contigs the follow a trim.seqs and carry on with the miseq SOP.

Hope that helps.

Bottom-line I rather use Mothur with a silva reference data base than qiime with greengenes.

If you're looking for a good reference on the QIIME full workflow process and detail on all the options here's an extensive chapter about it all: http://www.sciencedirect.com/science/article/pii/B9780124078635000198?#

QIIME also has lots of downstream options for processing.

Hi everyone,

I have a similar issue/problem. I'm a PhD student at the end of funding so time is not on my side and there's not much time I can devote to setting up new methods in the lab.

However, I am the first in our lab to perform next generation sequencing (Illumina Miseq). The samples are from anaerobic digestor. As such there is no one in my department that can help with any of the analysis. I have managed to install Mothur and QIIME myself but decided to start with Mothur as some people though this would be better? However, as I have paired end data (R1 and R2) like Sunils question. I am at a loss as to the first steps of process before I can follow the Mothur SOP?

(Our samples were sent to Mr. DNA so I don't necessarily have to work with the raw files as I have the OTU data etc. but I would like to do it especially for the advanced analysis and diversity indices)

Any help is greatly appreciated,

Ciara

Hey Ciara,

If you have OTU and taxonomy data, better to use the R for all other analysis. I am always using R for all kind of analysis on OTU matrix.

Rarefaction analysis

Ordination analysis

GLM analysis

and manu more option available depending upon your research question.

Believe me its really a powerful and user controlled tool as compare to all other (such as QIIME and MOTHUR)

In my case the major problem was mix orientation of reads in both R1 and R2 file and uneven tag size. Both qiime and MOTHUR doesnt handle this problem so far. But thank you all people for suggestion. Now problem has been solved.

Ah I did use R to produce heat maps but it was suggested to me to use Mothur for the other analysis starting from scratch. I need to look at shared OTU's, rarefaction curves, PCA plots, representative sequences and identifying shared sequences among different sequences sets etc. Perhaps I will try these in R first, I am still learning the basics of it but I did manage to produce nice heat maps from the data with it so hopefully the others won't be too difficult. Thanks so much for the advice.

Some suggestions...

For shared OTU and share sequences analysis you can try EstimateS and G-test (by QIIME)

For Rarefaction analysis you can follow Ugland et al procedure implemented in R

For PCA - R

I haven't try Heat-Maps specially in R (Yes, was trying to learn), could you please share me the protocol or script you used for heat map analysis in R

my email - [email protected]

OK if you have R1 and R2 this is what you need to do.

make an oligos file

type(assuming you are having a four core processor machine, and ur oligos file is named test.oligos)

make.contigs(ffastq=r1.fastq, rfastq=r2.fastq, oligos=test.oligos, processors=4)

from here on follow the mothur miseq SOP from step 2 onwards by making changes to match the resulting file names for your data.

Good luck.

Cheers

*refs

http://www.mothur.org/wiki/Make.contigs

http://www.mothur.org/wiki/MiSeq_SOP

http://www.mothur.org/wiki/Oligos_File

I have not used this one...but in case you dont want to try learn R, here is a possible alternative

http://vamps.mbl.edu/overview.php

Hello Jay,

So in MOTHUR where you will be able to the quality control if using make.contigs. in trim.seqs it was possible to do for 454 data.

Yes..deltaq option to be used

Qiime and Mothur are really powerful if used in the correct way... I'd suggest you to follow one of the many tutorial available on the web (http://qiime.org/1.6.0/tutorials/illumina_overview_tutorial.html) and easily you'll get great results. I also advice to use R especially Bioconductor packages like "phyloseq" (http://www.bioconductor.org/packages/2.12/bioc/html/phyloseq.html) that with the "ggplots" engine will help you making very high quality figures of e.g. heatmaps and ordination graphs and multivariate statistical analysis for community ecology data.

Hope it helps,

Gian

Hi Sunil,

I am in the process of making a protocol for the lab but I followed this tutorial http://www.molecularecologist.com/2013/08/making-heatmaps-with-r-for-microbiome-analysis/

and this book Instant Heatmaps In R (by Sebastian Rashka) he's on ResearchGate. My protocol will be more informative to Illumina Miseq data but I still have to finish it. In the meantime maybe try the protocol mentioned and email me at [email protected] if you get stuck? I'm happy to help!

Did you use the OTU table for analysis in R for regression etc? I'm reading books and tutorials and R but can't really find anything relevant to my data type as of yet! What is the name of the Ugland paper? Thank you.

Gian are there any tutorials/examples for using these analysis from phyloseq and ggplots for OTU data?

Thanks,

Ciara

Hey Caira here is the paper and scipt is online

Ugland KI, Gray JS, Ellingsen KE. 2003. The species–accumulation curve and estimation of species richness. Journal of Animal Ecology 72(5): 888-897.

Here is the rarefaction script

http://www.jennajacobs.org/R/rarefaction.txt

Hello jay,

deltaq option is for overlap base selection in make.contigs.

But i didnt find option for quality control of whole sequence after joining?

Regards

Sunil

by qc if you mean, removing homopolymers and short sequences or sequences with NNN then trim.seqs is what you are looking for.

if you want point and click for some visualisation this might be an option

http://www.ncbi.nlm.nih.gov/pubmed/24021386

Bioinformatics. 2013 Dec 1;29(23):3100-1. doi: 10.1093/bioinformatics/btt526. Epub 2013 Sep 10.

Explicet: graphical user interface software for metadata-driven management, analysis and visualization of microbiome data.

All my dataset is from illumina miseq. For the quality control, I thought you can try prinseq or cutadapt, both are good. For the figure, maybe R language is much more useful.

Working on MiSeq PE data, I have been using Trimmomatic to do the quality filtering. I then went into MOTHUR to assemble the forward an reverse paired reads from Trimmomatic using the "make.contigs" command. This will turn your fastq files into fasta files as well as assemble paired reads. If you have several samples/libraries, you should also make a group file using the "make.group" command. Then remember to use it in all commands that offer this parameter.

You can then merge all your fasta files using "cat" (in shell command promt, not mothur) as they will be indentified in subsequent analyses through your group file. I guess you could also use the "merge.files" command in MOTHUR, but I found it easier to use cat.

After this you can follow the MiSeq SOP proposed on the MOTHUR website.

Its usually a run with low phix spike or using the old software on miseq.

Mothur cannot help much with noisy data , but there might be some ways around it, if you really need to get some data out of it.

I think you should follow the UPARSE pipeline right from raw seq reads to otu_table. Its great for MiSeq data.  Advantage is it pools all the quality filtered reads into a single file before going for dereplication.