August 2011 - Mg-rast - mailman3.cels.anl.gov

Re: [mg-rast] MG-rast suggestion
by Folker Meyer 08 Aug '11

08 Aug '11

Maude: good idea. we will add that to the list of features for a future release. I suspect you'd like a linear recruitment plot for multiple reference genomes? Best, Folker On Aug 8, 2011, at 4:19 PM, Maude David wrote: > Hi, > > I use MG-RAST and I have one suggestion : that would be great to be able to select several reference genomes at once when we do the recruitment plot and we generate a table. because now we have to select and generate those table one by one to find the number of read matching with the microbes we are interested in. > > Thanks, > > Maude > > -- > Maude M. David > Postdoctoral researcher > tel: (510) 486-6154 > Departement of Ecology, Earth Sciences Division > Lawrence Berkeley National Laboratory > 1 Cyclotron Road, MS 70A-3317, Berkeley, CA 94720 > MMDavid(a)lbl.gov >

1 0

MG-rast suggestion
by Maude David 08 Aug '11

08 Aug '11

Hi, I use MG-RAST and I have one suggestion : that would be great to be able to select several reference genomes at once when we do the recruitment plot and we generate a table. because now we have to select and generate those table one by one to find the number of read matching with the microbes we are interested in. Thanks, Maude -- Maude M. David Postdoctoral researcher tel: (510) 486-6154 Departement of Ecology, Earth Sciences Division Lawrence Berkeley National Laboratory 1 Cyclotron Road, MS 70A-3317, Berkeley, CA 94720 MMDavid(a)lbl.gov

1 0

Re: [mg-rast] uploading Illumina pair-end data
by William Trimble 08 Aug '11

08 Aug '11

The best practive depends on whether you expect the reads to overlap or not. What is the ratio of your read length to your expected insert size? If the reads do not overlap, you should concatenate the FASTQ files and send the merged file to MG-RAST. If the reads do overlap, you should merge the mate pairs (for instance, Hammingaligner from illumina does this) and upload the merged reads. William Trimble --for the MG-RAST team. On Mon, Aug 8, 2011 at 11:05 AM, <David.beighton(a)kcl.ac.uk> wrote: > How should we upload illumina pair-end data? > Both fastq files? > Cheers David. >

1 0

uploading Illumina pair-end data
by David.beighton＠kcl.ac.uk 08 Aug '11

08 Aug '11

How should we upload illumina pair-end data? Both fastq files? Cheers David.

1 0

annotation files
by William Trimble 08 Aug '11

08 Aug '11

> I have been trying to download annotation information containing read id, hit details (evalue, length etc) and annotated organism/protein. I can only either get the read and organism/protein OR read id and hist details but not both in one file. Also, average for multiple reads is given instead of listing all the reads in the tab file, which I don't know how to go around... MG-RAST v3 doesn't assemble this information into one file, sorry. Per-read information is available by sending (subsets) of reads to the workbench and downloading them from there. > Ideally, I would öike to download the full analysis directly but I haven't been able to understand the files available there either. Is the pipeline and files available described somewhere? The formats of the files that come from the annotation pipeline are described here: http://blog.metagenomics.anl.gov/howto/downloads/ > I was also quite surprised to notice changes in the results for public datasets (mouse gut metagenomes) from last time I checked, about 2-3 months ago. I have saved the numbers and now they are different... What could be the reason for that...? Perhaps the analysis was not complete yet even though it was available for viewing....? >From time to time we make revisions to the software that presents the annotation results. In the past three months we had a revision that changed the way the similarity results were ranked and we had a revision to avoid some double-counting. These may have changed the numbers that you saw. I hope this helps. William Trimble --for the MG-RAST team

1 0

Re: [mg-rast] annotation files
by William Trimble 08 Aug '11

08 Aug '11

> I have been trying to download annotation information containing read id, hit details (evalue, length etc) and annotated organism/protein. I can only either get the read and organism/protein OR read id and hist details but not both in one file. Also, average for multiple reads is given instead of listing all the reads in the tab file, which I don't know how to go around... MG-RAST v3 doesn't assemble this information into one file, sorry. Per-read information is available by sending (subsets) of reads to the workbench and downloading them from there. > Ideally, I would öike to download the full analysis directly but I haven't been able to understand the files available there either. Is the pipeline and files available described somewhere? The formats of the files that come from the annotation pipeline are described here: http://blog.metagenomics.anl.gov/howto/downloads/ > I was also quite surprised to notice changes in the results for public datasets (mouse gut metagenomes) from last time I checked, about 2-3 months ago. I have saved the numbers and now they are different... What could be the reason for that...? Perhaps the analysis was not complete yet even though it was available for viewing....? >From time to time we make revisions to the software that presents the annotation results. In the past three months we had a revision that changed the way the similarity results were ranked and we had a revision to avoid some double-counting. These may have changed the numbers that you saw. I hope this helps. William Trimble --for the MG-RAST team

1 0

annotation files
by Katarzyna.Zaremba＠dieppe-1.mcs.anl.gov 08 Aug '11

08 Aug '11

Hi, I have been trying to download annotation information containing read id, hit details (evalue, length etc) and annotated organism/protein. I can only either get the read and organism/protein OR read id and hist details but not both in one file. Also, average for multiple reads is given instead of listing all the reads in the tab file, which I don't know how to go around... Ideally, I would �ike to download the full analysis directly but I haven't been able to understand the files available there either. Is the pipeline and files available described somewhere? I was also quite surprised to notice changes in the results for public datasets (mouse gut metagenomes) from last time I checked, about 2-3 months ago. I have saved the numbers and now they are different... What could be the reason for that...? Perhaps the analysis was not complete yet even though it was available for viewing....? thanks, Kasia

1 0

Re: [mg-rast] MG-RAST upload of 16S pyrosequencing reads through QIIME
by Andreas Wilke 05 Aug '11

05 Aug '11

Hi Marcus, thank you for your patience. a) qiime is using an old version to submit data to MG-RAST which is deprecated. This was built for our old MG-RAST Version 2. In this case a project ID is just free text. b) we have an new api to submit metagenomes to mg-rast and working with qiime to include it in there next release. In the meantime I'll fix the script so that you can upload your metagenomes and add it to your project. If you have a title for your project and a short description I can create the project for you and send you the ID. Do you have metadata too ? Andreas On Aug 5, 2011, at 5:20 AM, Marcus Claesson wrote: > Hi Andreas, > > Have you had any luck with finding out if I can use the Qiime submission > script? > > Many thanks, > Marcus > > > > On Fri, 2011-07-29 at 11:14 -0500, Andreas Wilke wrote: >> Dear Marcus , >> >> >> I have to check the version of the submission script you are using , >> we will get back to you as soon as possible. >> >> >> Andreas >> >> >> >> >> On Jul 29, 2011, at 11:01 AM, Marcus Claesson wrote: >> >>> Dear List, >>> >>> Is it currently possible to upload fasta files of >>> (barcode/primer-stripped) 16S pyrosequencing reads to my existing >>> MG-RAST account using this script: >>> http://qiime.sourceforge.net/scripts/submit_to_mgrast.html?highlight=submit… >>> >>> I have just generated an authorisation key but don't have the >>> Project >>> ID. How can I find that? >>> >>> Many thanks, >>> Marcus >>> >>> >> >> Andreas Wilke >> Argonne National Laboratory >> Mathematics and Computer Science Division >> Bldg. 240 , 4F8 >> 9700 South Cass Avenue >> Argonne, IL 60439 >> (630) 252-3190 (phone) >> >> (630) 252-5986.(fax) >> >> > Andreas Wilke Argonne National Laboratory Mathematics and Computer Science Division Bldg. 240 , 4F8 9700 South Cass Avenue Argonne, IL 60439 (630) 252-3190 (phone) (630) 252-5986.(fax)

1 0

Re: [mg-rast] problem with very large sff file uploaded and processed incorrectly
by William Trimble 05 Aug '11

05 Aug '11

So http://metagenomics.anl.gov/metagenomics.cgi?page=MetagenomeOverview&metage… I see only 4% of your sequence data made it through quality filtering. Quality filtering probably took too much from this one; you should try running your data again either using dynamic trimming or with a less stringent cutofff (say, Q=20, N=5). William Trimble --for the MG-RAST team On Thu, Aug 4, 2011 at 12:23 PM, Alison Murray <Alison.Murray(a)dri.edu> wrote: > Thanks William. I do think the build that you refer to is the most recent > one (as far as I know at least). Not sure what other organisms would be > good to have - is C. elegans there already? I guess that could also be > relevant for some studies. If you add the Ciona - then I'll try and rerun > the analysis. > > I just checked to see where my file was in the pipeline - and can't believe > my large data set has actually finished! I guess the erroneous file > submitted earlier in July must have "crashed" - such that I didn't figure > out it was screwed up until over 3 weeks after it was submitted! In the > end... very happy I can see what we get. I did "loose" a ton of sequences > following QC - I used a phred score cutoff of 25 (and I think 5 low quality > bases per read) - that's the "main" thing I did there for quality. In your > experience with metagenomic data, is that too conservative for 454 data? I > didn't really think so -but am wondering now whether I should rerun the > analysis with a lower score. > > Thanks for the help, > Alison > > > > ----- Original Message ----- > From: William Trimble <wltrimbl(a)gmail.com> > Date: Thursday, August 4, 2011 7:20 am > Subject: Re: [mg-rast] problem with very large sff file uploaded and > processed incorrectly > To: Alison Murray <Alison.Murray(a)dri.edu>, mg-rast(a)mcs.anl.gov > >> We can add genomes to the list of model organisms >> for which MG-RAST provides subtractive read filters. >> >> Is the Ciona intestinalis v2 genome assembly, from 2005, >> the most recent/suitable genome ? >> http://genome.jgi-psf.org/Cioin2/Cioin2.download.ftp.html >> >> Are there any other organisms that you can suggest >> we add to the list of organisms to screen against? >> Dog, chicken, and sea squirt are already on the list. >> >> William Trimble >> --for the MG-RAST team >> >> >> On Wed, Aug 3, 2011 at 11:31 AM, William Trimble >> <wltrimbl(a)gmail.com> wrote: >> > MG-RAST has a sequence-subtracting filter that >> > can remove sequences matching >> > human, cow, mouse, fruit fly, arabidopsis, or E. Coli >> > before the sequences enter the similarity pipeline. >> > >> > At present we don't have a function to screen against >> > Ciona intestinalis, sorry. >> > >> > William Trimble >> > --for the MG-RAST team >> > >> > >> > On Wed, Aug 3, 2011 at 11:17 AM, Alison Murray >> <Alison.Murray(a)dri.edu> wrote: >> >> >> >> Thanks William. >> >> >> >> Do you know if there's any way to add the Ciona intestinalis >> genome to use >> >> as a reference genome to subtract sequences from this whole >> data set? (I'm >> >> not actually sure at this point whether it would even work >> well for >> >> subtraction - but at least it's in the right phylum as my >> host organism). >> >> >> >> I guess I'll find out how much contamination there is >> following the analysis >> >> - but it would be helpful to add not the top few model >> organisms (I'd think >> >> that Ciona would have been a choice actually). >> >> >> >> Cheers, >> >> Alison >> >> >> >> >> >> ----- Original Message ----- >> >> From: William Trimble <wltrimbl(a)gmail.com> >> >> Date: Wednesday, August 3, 2011 7:08 am >> >> Subject: Re: [mg-rast] problem with very large sff file >> uploaded and >> >> processed incorrectly >> >> To: Alison Murray <Alison.Murray(a)dri.edu>, mg-rast(a)mcs.anl.gov >> >> >> >>> We got 1.57 M sequences from the file you uploaded >> >>> last night, so it looks like it came through this time. >> >>> >> >>> It should go through the pipeline now. >> >>> >> >>> We'll have to delete the broken job for you. >> >>> >> >>> William Trimble >> >>> --for the MG-RAST team >> >>> >> >>> On Wed, Aug 3, 2011 at 8:56 AM, Alison Murray >> >>> <Alison.Murray(a)dri.edu> wrote: >> >>> > Hi William, >> >>> > >> >>> > I uploaded a new dataset to replace the one that ran >> >>> incorrectly this last >> >>> > month. "Info: Your upload will be processed as job number 29212" >> >>> > >> >>> > You indicated that you'd check the file quality - that would >> >>> be much >> >>> > appreciated. >> >>> > >> >>> > I selected the options to dereplicated this data set as well >> >>> as dynamic >> >>> > trimming - phred score of 25 of max of 5 low quality bases >> per read. >> >>> > >> >>> > I tried to see if I could select the genome of Ciona >> >>> intestinalis as a model >> >>> > genome to screen against, but it did not seem that this was >> >>> possible. If >> >>> > that is possible, that is the closest organism with a genome >> >>> sequence that >> >>> > I'm aware of that could be used to screen against - as >> this is a >> >>> > host-associated microbiome metagenome that we've sequenced here. >> >>> > >> >>> > Also - if possible could you delete the Norsel3- >> >>> S_adreanum_MicrobiomeMG from >> >>> > my analysis cart (or can I delete it)? >> >>> > I'll be away for the next 2 hrs, then back around 9:15 am >> >>> PST. Much >> >>> > appreciate your help. >> >>> > >> >>> > >> >>> > >> >>> > ----- Original Message ----- >> >>> > From: William Trimble <wltrimbl(a)gmail.com> >> >>> > Date: Monday, August 1, 2011 12:00 pm >> >>> > Subject: Re: [mg-rast] problem with very large sff file >> >>> uploaded and >> >>> > processed incorrectly >> >>> > To: Alison Murray <Alison.Murray(a)dri.edu>, mg- >> rast(a)mcs.anl.gov>>> > >> >>> >> Both of the two recommended approaches (concatenated >> >>> >> FASTQ or separate SFF files) keep the per-base quality data. >> >>> >> >> >>> >> The FASTA format, on the other hand, doesn't retain the >> >>> >> quality scores for each read. >> >>> >> >> >>> >> William Trimble >> >>> >> --for the MG-RAST team >> >>> >> >> >>> >> On Mon, Aug 1, 2011 at 1:53 PM, Alison Murray >> >>> >> <Alison.Murray(a)dri.edu> wrote: >> >>> >> > ahh, >> >>> >> > >> >>> >> > OK. >> >>> >> > >> >>> >> > Is there information in the sff file that will be lost >> >>> >> (quality info?) that >> >>> >> > would make it better to run the two files separately? I might >> >>> >> also have the >> >>> >> > fastaq files. Or if not - can figure out how to >> prepare them. >> >>> >> > >> >>> >> > In terms of analysis - it's more attractive to have the whole >> >>> >> lot analyzed >> >>> >> > together - but I'm aware that quality is a critical issue. >> >>> >> What do you >> >>> >> > recommend then for the highest quality anlaysis? >> >>> >> > >> >>> >> > -Alison >> >>> >> > >> >>> >> > >> >>> >> > ----- Original Message ----- >> >>> >> > From: William Trimble <wltrimbl(a)gmail.com> >> >>> >> > Date: Monday, August 1, 2011 11:49 am >> >>> >> > Subject: Re: [mg-rast] problem with very large sff file >> >>> >> uploaded and >> >>> >> > processed incorrectly >> >>> >> > To: Alison Murray <Alison.Murray(a)dri.edu>, mg- >> rast(a)mcs.anl.gov>>> >> > >> >>> >> >> Oh, that's probably the reason, then. Concatenating >> >>> >> >> SFF files requires programs that deliberately write >> >>> >> >> SFF files. >> >>> >> >> >> >>> >> >> In that case, you could either send us the two SFF files >> >>> >> >> separately and run them as two separate jobs or >> >>> >> >> convert both runs to FASTQ and upload the concatenated >> >>> >> >> FASTQ. Either one will work with MG-RAST. >> >>> >> >> >> >>> >> >> William Trimble >> >>> >> >> --for the MG-RAST team >> >>> >> >> >> >>> >> >> On Mon, Aug 1, 2011 at 1:40 PM, Alison Murray >> >>> >> >> <Alison.Murray(a)dri.edu> wrote: >> >>> >> >> > OK, >> >>> >> >> > >> >>> >> >> > will start that shortly. I did combine two sff files >> >>> into the >> >>> >> >> one that I >> >>> >> >> > uploaded; I wonder if that caused the problem? I >> wanted to >> >>> >> >> join all the >> >>> >> >> > sequence I have from this data set into one analysis >> - >> >>> we had >> >>> >> >> a small set of >> >>> >> >> > data from the titration runs, and then the big main >> data set. >> >>> >> >> The file that >> >>> >> >> > I have to upload is the combined one (I concatenated >> the two >> >>> >> >> sff files). >> >>> >> >> > Actually - if you think that's a problem, let me know >> >>> and I'll >> >>> >> >> do them >> >>> >> >> > separately. It's all the same library though so >> would like >> >>> >> >> the analysis >> >>> >> >> > ideally to consider all the sequence files ran in >> these two >> >>> >> >> batches from the >> >>> >> >> > same data set. >> >>> >> >> > >> >>> >> >> > -Alison >> >>> >> >> > >> >>> >> >> > >> >>> >> >> > >> >>> >> >> > ----- Original Message ----- >> >>> >> >> > From: William Trimble <wltrimbl(a)gmail.com> >> >>> >> >> > Date: Monday, August 1, 2011 11:35 am >> >>> >> >> > Subject: Re: [mg-rast] problem with very large sff file >> >>> >> >> uploaded and >> >>> >> >> > processed incorrectly >> >>> >> >> > To: Alison Murray <Alison.Murray(a)dri.edu> >> >>> >> >> > >> >>> >> >> >> It appears that we don't have an intact copy >> >>> >> >> >> of your dataset. >> >>> >> >> >> >> >>> >> >> >> Yes, you should just upload the file, let us >> >>> >> >> >> know, and if it looks ok you can create a new job. >> >>> >> >> >> >> >>> >> >> >> William Trimble >> >>> >> >> >> --for the MG-RAST team >> >>> >> >> >> >> >>> >> >> >> >> >>> >> >> >> On Mon, Aug 1, 2011 at 1:25 PM, Alison Murray >> >>> >> >> >> <Alison.Murray(a)dri.edu> wrote: >> >>> >> >> >> > Hi William, >> >>> >> >> >> > >> >>> >> >> >> > The numbers are slightly different from what I see. >> >>> I think >> >>> >> >> >> that I >> >>> >> >> >> > compressed a 7,539,547 SFF file - such that the one >> >>> that was >> >>> >> >> >> uploaded was >> >>> >> >> >> > 3,651,001. (File name Norsel3MG_all.sff). >> >>> >> >> >> > >> >>> >> >> >> > Should I go ahead and try and upload through the MGRAST >> >>> >> >> >> browser this >> >>> >> >> >> > compressed file? >> >>> >> >> >> > >> >>> >> >> >> > From what you suggest - I think the answer is >> yes, but >> >>> >> am not >> >>> >> >> >> sure of the >> >>> >> >> >> > differences in file size. >> >>> >> >> >> > >> >>> >> >> >> > -Alison >> >>> >> >> >> > >> >>> >> >> >> > >> >>> >> >> >> > >> >>> >> >> >> > >> >>> >> >> >> > ----- Original Message ----- >> >>> >> >> >> > From: William Trimble <wltrimbl(a)gmail.com> >> >>> >> >> >> > Date: Monday, August 1, 2011 11:18 am >> >>> >> >> >> > Subject: Re: [mg-rast] problem with very large >> sff file >> >>> >> >> >> uploaded and >> >>> >> >> >> > processed incorrectly >> >>> >> >> >> > To: Alison Murray <Alison.Murray(a)dri.edu>, mg- >> >>> >> rast(a)mcs.anl.gov>> >> > >> >>> >> >> >> >> Thank you for letting us know you're having trouble. >> >>> >> >> >> >> >> >>> >> >> >> >> You uploaded a compressed 7 Gb SFF file, but >> >>> >> >> >> >> the file had a problem that caused us to find >> >>> >> >> >> >> only 74 reads in it. >> >>> >> >> >> >> >> >>> >> >> >> >> This problem was probably caused by an error >> during the >> >>> >> >> >> >> upload process. >> >>> >> >> >> >> >> >>> >> >> >> >> The SFF file that we got was 7720495172 bytes long. >> >>> >> >> >> >> does this match the size of the file that you sent? >> >>> >> >> >> >> >> >>> >> >> >> >> You will need to upload the compressed SFF file again. >> >>> >> >> >> >> We can then check to make sure it is complete. >> >>> >> >> >> >> >> >>> >> >> >> >> William Trimble >> >>> >> >> >> >> --for the MG-RAST team >> >>> >> >> >> >> >> >>> >> >> >> >> >> >>> >> >> >> >> On Mon, Aug 1, 2011 at 12:40 PM, Alison Murray >> >>> >> >> >> >> <Alison.Murray(a)dri.edu> wrote: >> >>> >> >> >> >> > Hi there, >> >>> >> >> >> >> > >> >>> >> >> >> >> > I uploaded a very large file - over 3.6GB that was >> >>> a sff >> >>> >> >> >> file (name: >> >>> >> >> >> >> > Norsel3-S_adreanum_MicrobiomeMG) a few weeks >> ago - >> >>> >> this was >> >>> >> >> >> >> submitted for >> >>> >> >> >> >> > analysis - and today I checked, and it was >> >>> "available for >> >>> >> >> >> analysis".>> > However, it appears that only a very small >> >>> >> >> >> portion of the >> >>> >> >> >> >> sequences were >> >>> >> >> >> >> > actually uploaded/processed. I'm wondering if >> >>> >> somehow the >> >>> >> >> >> >> criteria for QC >> >>> >> >> >> >> > was too strict? Or if I crashed the server??? >> Really>>> >> >> have no >> >>> >> >> >> >> clue as to why >> >>> >> >> >> >> > it didn't work - but definitely a problem >> here. This >> >>> >> is a >> >>> >> >> >> >> huge metagenome >> >>> >> >> >> >> > data set - generated with 454. >> >>> >> >> >> >> > >> >>> >> >> >> >> > Would appreciate it if someone could help me with >> >>> >> this - as >> >>> >> >> >> >> it's literally >> >>> >> >> >> >> > be 3 weeks waiting to then see that something >> >>> didn't work >> >>> >> >> >> >> right - in the >> >>> >> >> >> >> > upload/processing settings were mostly the default >> >>> >> >> settings as >> >>> >> >> >> >> far as I >> >>> >> >> >> >> > recall. I uploaded the raw, non-assembled data set. >> >>> >> (supposed>> >> >> to be 865 MB >> >>> >> >> >> >> > of raw sequence) >> >>> >> >> >> >> > >> >>> >> >> >> >> > Thanks for the help! >> >>> >> >> >> >> > >> >>> >> >> >> >> > Alison >> >>> >> >> >> >> > >> >>> >> >> >> >> > >> >>> >> >> >> >> > >> >>> >> >> >> >> > Alison E. Murray, Ph.D.// >> >>> >> >> >> >> > Associate Research Professor// >> >>> >> >> >> >> > Desert Research Institute// >> >>> >> >> >> >> > 2215 Raggio Parkway// >> >>> >> >> >> >> > Reno, NV 89512// >> >>> >> >> >> >> > >> >>> >> >> >> >> > Tel: 775 673 7361// >> >>> >> >> >> >> > FAX: 775 673 7635// >> >>> >> >> >> >> > http://www.dri.edu/People/alison/ >> >>> >> >> >> >> > >> >>> >> >> >> >> > >> >>> >> >> >> > >> >>> >> >> >> > Alison E. Murray, Ph.D.// >> >>> >> >> >> > Associate Research Professor// >> >>> >> >> >> > Desert Research Institute// >> >>> >> >> >> > 2215 Raggio Parkway// >> >>> >> >> >> > Reno, NV 89512// >> >>> >> >> >> > >> >>> >> >> >> > Tel: 775 673 7361// >> >>> >> >> >> > FAX: 775 673 7635// >> >>> >> >> >> > http://www.dri.edu/People/alison/ >> >>> >> >> >> > >> >>> >> >> >> > >> >>> >> >> > >> >>> >> >> > Alison E. Murray, Ph.D.// >> >>> >> >> > Associate Research Professor// >> >>> >> >> > Desert Research Institute// >> >>> >> >> > 2215 Raggio Parkway// >> >>> >> >> > Reno, NV 89512// >> >>> >> >> > >> >>> >> >> > Tel: 775 673 7361// >> >>> >> >> > FAX: 775 673 7635// >> >>> >> >> > http://www.dri.edu/People/alison/ >> >>> >> >> > >> >>> >> >> > >> >>> >> > >> >>> >> > Alison E. Murray, Ph.D.// >> >>> >> > Associate Research Professor// >> >>> >> > Desert Research Institute// >> >>> >> > 2215 Raggio Parkway// >> >>> >> > Reno, NV 89512// >> >>> >> > >> >>> >> > Tel: 775 673 7361// >> >>> >> > FAX: 775 673 7635// >> >>> >> > http://www.dri.edu/People/alison/ >> >>> >> > >> >>> >> > >> >>> > >> >>> > Alison E. Murray, Ph.D.// >> >>> > Associate Research Professor// >> >>> > Desert Research Institute// >> >>> > 2215 Raggio Parkway// >> >>> > Reno, NV 89512// >> >>> > >> >>> > Tel: 775 673 7361// >> >>> > FAX: 775 673 7635// >> >>> > http://www.dri.edu/People/alison/ >> >>> > >> >>> > >> >> >> >> Alison E. Murray, Ph.D.// >> >> Associate Research Professor// >> >> Desert Research Institute// >> >> 2215 Raggio Parkway// >> >> Reno, NV 89512// >> >> >> >> Tel: 775 673 7361// >> >> FAX: 775 673 7635// >> >> http://www.dri.edu/People/alison/ >> >> >> >> >> > > > Alison E. Murray, Ph.D.// > Associate Research Professor// > Desert Research Institute// > 2215 Raggio Parkway// > Reno, NV 89512// > > Tel: 775 673 7361// > FAX: 775 673 7635// > http://www.dri.edu/People/alison/ > >

1 0

Re: [mg-rast] Question about number of hits
by William Trimble 05 Aug '11

05 Aug '11

About 300,000 of the reads are annotated for your untrimmed sample. The fact that only half were annotated is a bit of a puzzle. The minimum alignment lengths are the defaults of blat. For the data that contain the primer, we don't have a way to mask the alignment if it includes the primer. Your data might benefit from a search against a smaller database than MG-RAST uses. Since you have low annotation fractions, a database with only V6 sequences might have better sensitivity. I hope this helps. William Trimble --for the MG-RAST team On Fri, Aug 5, 2011 at 12:51 PM, Turki Abujamel <abujamel.t(a)gmail.com> wrote: > Hi William, > > Even when including the amplification primers, the number of hits still low. You could try this sample : 4469103.3, before trimming; 4469395.3, after trimming. But my concern is, what is the minimum red length cutoff in the program? Because I am worried that the software might align to the primer sequence, and that could be the reason for the high hit number. > > Thanks again, > Turki > > On 2011-08-05, at 1:24 PM, William Trimble wrote: > >> I can confirm your dataset has 600,000 reads, that it is >> amplicon, and that the sum of the abundance >> counts from the SSU annotations is only about >> 100,000. >> >> It looks like the step that is failing is the similarity >> search step; only 200 of the 750 clusters have >> hits to the database. >> >> I suspect that the fraction identified would be >> higher if you included the amplification primers. >> Though they don't have information specific >> to the species in your sample, they will give >> our general-purpose similarity engine >> confidence that your reads really do align to >> the bacterial V6 region (which it doesn't seem >> to be doing for most of your reads). >> >> Low fraction of reads identified is something we >> see sometimes, usually with reads that are very >> short. >> >> I hope this helps. >> >> William Trimble >> --for the MG-RAST team >> >> >> On Fri, Aug 5, 2011 at 11:46 AM, Turki Abujamel <abujamel.t(a)gmail.com> wrote: >>> Thank you william for your fast reply. >>> >>> the sample id is: 4469385.3 >>> >>> Regards, >>> Turki >>> >>> On 2011-08-05, at 12:41 PM, William Trimble wrote: >>> >>>> The RNA-similarity part of MG-RAST performs clustering >>>> of the reads at 97 % identity, performs the similarity search >>>> on a representative of the each cluster, and multiplies the >>>> similarity results by the size of each cluster to estimate >>>> abundance. We don't sample the reads randomly to >>>> reduce compute load. >>>> >>>> That said, the low fraction identified may be a side effect >>>> of the short reads. >>>> >>>> Can you give us the MG-RAST ID of the sample you're >>>> thinking about and we can look at it in detail? >>>> >>>> William Trimble >>>> --for the MG-RAST team >>>> >>>> On Fri, Aug 5, 2011 at 11:20 AM, Turki Abujamel <abujamel.t(a)gmail.com> wrote: >>>>> Hi, >>>>> I have Illumina sequences for bacterial V6 region of 16S rRNA gene. I >>>>> noticed that the number of hits is much lower than the number of reads. For >>>>> example, in one of the files, I have about 600,000 sequences with an average >>>>> read length of 74 nucleotides and lowest reads length of about 65 >>>>> nucleotides. I noticed when using the table to see the assigned bacterial >>>>> species using RDP as annotation source that there is only about 100,000 hits >>>>> even when not using any cutoffs. It was also the same when repeating the >>>>> process at different phylogenetic levels. Does this mean that the program >>>>> randomly pick a percentage of the sequences and use them for assigning >>>>> taxonomy in order to reduce the computer power needed? My sequences are high >>>>> quality sequences. They were originally filtered based on quality score, >>>>> presence of amplification primers, and 50 bp read length cutoff. Originally >>>>> the read length was 120, but the size now is ~ 74 after amplification >>>>> primers trimming. >>>>> Regards, >>>>> Turki Abujamel >>>>> >>>>> B.Sc., M.Sc., >>>>> Ph.D Candidate >>>>> Dr. Stintzi's Lab >>>>> Ottawa Institute of Systems Biology, BMI department, >>>>> Faculty of Medicine, University of Ottawa >>>>> 451, Smyth Road, Ottawa, Ontario, K1H8M5. >>>>> e-mail: abujamel.t(a)gmail.com >>>>> Phone: +1 (613) 613-562-5459 extension 8220 >>>>> Cell Phone: +1 (613) 796-6644 >>> >>> > >

1 0