January 2008 - Mg-rast - mailman3.cels.anl.gov

Re: mg-rast guest account pw-changes
by Daniel Paarmann 31 Jan '08

31 Jan '08

Yes, that is a known issue. We already introduced a right for that and will update the account management accordingly. Daniel On 31.01.2008, at 14:35, Folker Meyer wrote: > HI, > > I have just realized that I can change the PW for the guest account... > > Maybe we want to eliminate the menu entry for manage for the guest > account? > > Folker >

1 0

mg-rast guest account pw-changes
by Folker Meyer 31 Jan '08

31 Jan '08

HI, I have just realized that I can change the PW for the guest account... Maybe we want to eliminate the menu entry for manage for the guest account? Folker

1 0

mg-rast another update to genomemeta
by Robert Olson 31 Jan '08

31 Jan '08

I think I may have finally gotten the fix to the genome meta stuff. needed to fsync() the file after writing - getting the locking right was mostly the fix, but NFS caching gets in the way for very heavy write loads. --bob

1 0

Re: mg-rast Fwd: Weirdness with the Mom samples
by Robert Olson 30 Jan '08

30 Jan '08

Is this the stuff where the the current object of your attention is stored in a cookie? if it is that mechanism would need to change... On Jan 30, 2008, at 1:04 PM, Rob Edwards wrote: > Just making sure everyone else is aware that this problem of having > multiple page views open simultaneously keeps cropping up. Is there > an easy fix? > > Rob > > > Begin forwarded message: > >> From: "Forest Rohwer" <frohwer(a)gmail.com> >> Date: January 30, 2008 11:01:06 AM PST >> To: "Florent Angly" <florent.angly(a)gmail.com> >> Cc: "Rebecca Vega Thurber" <rvegathurber(a)gmail.com>, "Matthew >> Haynes" <mhaynes(a)projects.sdsu.edu>, "Rob Edwards" >> <raedwards(a)gmail.com> >> Subject: Re: Weirdness with the Mom samples >> >> If you logout and login again, it fixes that problem. Rob is aware >> of this. >> FLR >> >> On Jan 30, 2008 10:44 AM, Florent Angly <florent.angly(a)gmail.com> >> wrote: >> Hi Becky, >> >> Just so you know:iIt has already happened in the past that >> metagenomes >> have no 16sRDNA hits but... >> >> I tried to look at the 6 metagenomes and... >> >> In the taxonomy, what I see is that the 6 metagenomes appear >> exactly the >> same (16S rDNA and all sequences). That can NOT be true. On the main >> metagenomic SEED Viewer page, click on This Organism > Overview. The >> page is pretty much blank. There must be something wrong with the >> metagenomic SEED. Can you confirm that Becky? >> >> We need to let these guys know if that' s buggy. >> >> Florent >> >> Rebecca Vega Thurber wrote: >> > So I keep finding odd discrepancies in the poo metagenomes and I am >> > starting to think that something is wrong with the Mom's data. I >> > think that maybe those samples didn't go through the pipeline in >> the >> > same way cause there are whole sections on the RASt that are >> blank for >> > them...like the 16S similarities. All the twins have lots of >> hits but >> > the Mom's have zero, like they were actually never run. Ideas? >> -Becky >> > >> > -- >> > Rebecca Vega Thurber >> > NSF Postdoctoral Fellow >> > San Diego State University >> > Biology Department >> > LS301 North Life Sciences >> > 5500 Campanile Drive >> > San Diego, CA 92182-4614 >> > (619) 594-1336 >> > (619) 594-5676 FAX >> > rvegathurber(a)gmail.com <mailto:[email protected]> >> >> >

1 0

mg-rast Fwd: Weirdness with the Mom samples
by Rob Edwards 30 Jan '08

30 Jan '08

Just making sure everyone else is aware that this problem of having multiple page views open simultaneously keeps cropping up. Is there an easy fix? Rob Begin forwarded message: > From: "Forest Rohwer" <frohwer(a)gmail.com> > Date: January 30, 2008 11:01:06 AM PST > To: "Florent Angly" <florent.angly(a)gmail.com> > Cc: "Rebecca Vega Thurber" <rvegathurber(a)gmail.com>, "Matthew > Haynes" <mhaynes(a)projects.sdsu.edu>, "Rob Edwards" > <raedwards(a)gmail.com> > Subject: Re: Weirdness with the Mom samples > > If you logout and login again, it fixes that problem. Rob is aware > of this. > FLR > > On Jan 30, 2008 10:44 AM, Florent Angly <florent.angly(a)gmail.com> > wrote: > Hi Becky, > > Just so you know:iIt has already happened in the past that metagenomes > have no 16sRDNA hits but... > > I tried to look at the 6 metagenomes and... > > In the taxonomy, what I see is that the 6 metagenomes appear > exactly the > same (16S rDNA and all sequences). That can NOT be true. On the main > metagenomic SEED Viewer page, click on This Organism > Overview. The > page is pretty much blank. There must be something wrong with the > metagenomic SEED. Can you confirm that Becky? > > We need to let these guys know if that' s buggy. > > Florent > > Rebecca Vega Thurber wrote: > > So I keep finding odd discrepancies in the poo metagenomes and I am > > starting to think that something is wrong with the Mom's data. I > > think that maybe those samples didn't go through the pipeline in the > > same way cause there are whole sections on the RASt that are > blank for > > them...like the 16S similarities. All the twins have lots of > hits but > > the Mom's have zero, like they were actually never run. Ideas? - > Becky > > > > -- > > Rebecca Vega Thurber > > NSF Postdoctoral Fellow > > San Diego State University > > Biology Department > > LS301 North Life Sciences > > 5500 Campanile Drive > > San Diego, CA 92182-4614 > > (619) 594-1336 > > (619) 594-5676 FAX > > rvegathurber(a)gmail.com <mailto:[email protected]> > >

1 0

mg-rast Re: failed jobs
by Jesse Zaneveld 29 Jan '08

29 Jan '08

Hi Rob, Thank you once again for your assistance, and I hope that errors in processing my sequences did not interfere with other users annotations. The sequences I was submitting are annotated intergenic regions extracted from KEGG genomes. For several genomes or plasmids, the percentage of non-coding regions based on the annotated gene boundaries seemed implausibly high (in some cases above 50% noncoding!), and we were suspicious that some unannotated coding sequences might be present in the putative intergenic regions, especially since some of these regions were quite long. I was hoping to use mg-rast as a relatively broad net to catch genomes or plasmids for which there are a large number of unannotated genes, so that they could be excluded from downstream analyses. We thought that some genes might be found by a similarity-based approach that would be missed by, for example, Markov Model based methods when the overall size of the genome is small (most of our examples are small plasmids). Thus, I was submitting files containing all intergenic regions for each of ~70 chromosomes or plasmids which each had >30% of their nucleotides annotated as noncoding based on the gene boundaries identified in KEGG. I had expected that if no genes were present I would just get a genbank file with no 'CDS' feature for those regions for which no similarities could be identified, as seems to occur for individual records within a file for which no similarities can be found. Are errors raised from files containing no sequences with significant similarity to known sequences the expected behavior, or is this something that might change in the future? I looked into your suggestion to concatenate sequences from several genomes into a single file and upload them in that way, however, because as far as I can tell the fasta label lines are not incorporated into the output genbank file, I don't think I would be able to determine which sequences in the final result came from which original genome using this approach. If we know that each of the genomes that failed at stage 4 did so because no regions with significant similarity to known coding sequences could be found, then that could be a solution, because it would have identified those genomes for which we can be reasonably sure that no unannotated genes lurk in the non-coding regions. However, to make this inference we would have to know that this was the only reason this particular set of genomes failed at stage 4, and using the web interface, I was unable to find a way to get an error message describing the cause of each error. If based on the error reports that you see on your end (or if there is some way to access the error messages directly that I missed), we can determine that all the failures at stage 4 were due to this cause then I think that would actually be sufficient information for this project. In any case, sorry for the trouble and thank you for your support and assistance. ~Jesse Zaneveld On Jan 28, 2008 3:26 PM, Rob Edwards <redwards(a)mcs.anl.gov> wrote: > Hi Jesse, > > Many of the small jobs that you are submitting to the mg-rast are > failing because there are no similarities being detected, and so > nothing downstream can compute. > > I wonder if the mg-rast is the most appropriate way for you to > analyze your data? Is there something specific that you are trying to > accomplish that we could help with? > > Alternatively you might try combining several sequences into one > larger file. > > Rob > On behalf of the mg-rast team >

1 0

Re: mg-rast Bug with micro-jobs
by Rob Edwards 28 Jan '08

28 Jan '08

Another thing that would help is if failing to read the meta.xml is not fatal. I.e. at the moment if there is an error in an xml file the jobs overview will not load and will throw an error. This blocks everyone from accessing the site. It would be better to flag those somehow (e.g. coloring the whole row of boxes red) so that you can see which jobs have problematic xml files. Rob On Jan 28, 2008, at 2:24 PM, Robert Olson wrote: > If someone updates the production FigKernelPackages/GenomeMeta.pm > we may get this fixed. I've been running with this new version over > on the new RAST without problem. > > On Jan 28, 2008, at 4:18 PM, Rob Edwards wrote: > >> and most are causing havoc, too, because the meta.xml file is not >> being locked properly! >> >> >> >> On Jan 28, 2008, at 1:39 PM, Robert Olson wrote: >> >>> There's a current flood of tiny jobs (one had 3 reads) in the >>> mgrast server. Some are failing during sims postprocess in this >>> code: >>> >>> sub handle_block >>> { >>> my($block, $n) = @_; >>> >>> my @b = sort { $a->[4] <=> $b->[4] } @$block; >>> >>> for (my $i = 0; $i < $n; $i++) >>> { >>> my($id1, $id2, $genome) = @{$b[$i]}; >>> >>> >>> My guess is that this is because the blocks are singletons: >>> >>> Looking through 1 hits >>> Can't use an undefined value as an ARRAY reference at /vol/ >>> metagenome-48-hour/FIGdisk.backend.prod/FIG/bin/ >>> summarize_taxonomy_by_blast line 226. >>> >>> >>> --bob >>> >> >

1 0

mg-rast failed jobs
by Rob Edwards 28 Jan '08

28 Jan '08

Hi Jesse, Many of the small jobs that you are submitting to the mg-rast are failing because there are no similarities being detected, and so nothing downstream can compute. I wonder if the mg-rast is the most appropriate way for you to analyze your data? Is there something specific that you are trying to accomplish that we could help with? Alternatively you might try combining several sequences into one larger file. Rob On behalf of the mg-rast team

1 0

Re: mg-rast Bug with micro-jobs
by Robert Olson 28 Jan '08

28 Jan '08

If someone updates the production FigKernelPackages/GenomeMeta.pm we may get this fixed. I've been running with this new version over on the new RAST without problem. On Jan 28, 2008, at 4:18 PM, Rob Edwards wrote: > and most are causing havoc, too, because the meta.xml file is not > being locked properly! > > > > On Jan 28, 2008, at 1:39 PM, Robert Olson wrote: > >> There's a current flood of tiny jobs (one had 3 reads) in the >> mgrast server. Some are failing during sims postprocess in this code: >> >> sub handle_block >> { >> my($block, $n) = @_; >> >> my @b = sort { $a->[4] <=> $b->[4] } @$block; >> >> for (my $i = 0; $i < $n; $i++) >> { >> my($id1, $id2, $genome) = @{$b[$i]}; >> >> >> My guess is that this is because the blocks are singletons: >> >> Looking through 1 hits >> Can't use an undefined value as an ARRAY reference at /vol/ >> metagenome-48-hour/FIGdisk.backend.prod/FIG/bin/ >> summarize_taxonomy_by_blast line 226. >> >> >> --bob >> >

1 0

Re: mg-rast Bug with micro-jobs
by Rob Edwards 28 Jan '08

28 Jan '08

and most are causing havoc, too, because the meta.xml file is not being locked properly! On Jan 28, 2008, at 1:39 PM, Robert Olson wrote: > There's a current flood of tiny jobs (one had 3 reads) in the > mgrast server. Some are failing during sims postprocess in this code: > > sub handle_block > { > my($block, $n) = @_; > > my @b = sort { $a->[4] <=> $b->[4] } @$block; > > for (my $i = 0; $i < $n; $i++) > { > my($id1, $id2, $genome) = @{$b[$i]}; > > > My guess is that this is because the blocks are singletons: > > Looking through 1 hits > Can't use an undefined value as an ARRAY reference at /vol/ > metagenome-48-hour/FIGdisk.backend.prod/FIG/bin/ > summarize_taxonomy_by_blast line 226. > > > --bob >

1 0