Thanks you too!! I will give it a try! On Wed, Jul 24, 2013 at 11:37 PM, Tim Triche, Jr. <tim.triche@gmail.com>wrote: > that is a very cool resource -- if it could be filtered the way GEO > metadata can be, it would be just about perfect > > thank you for the pointer (speaking for myself here), I am bookmarking it > for my own later use :-) > > best, > > --t > > > On Wed, Jul 24, 2013 at 7:03 PM, Levi Waldron <lwaldron.research@gmail.com> > wrote: > >> Lina, >> >> you might try using insilicodb.org to facilitate the curation you need >> to do, then import ExpressionSets with your curated metadata from there. >> >> -Levi >> >> >> On Wed, Jul 24, 2013 at 4:06 PM, Lina Thomas <linadth6@gmail.com> wrote: >> >>> Well thanks a lot! In some way that is going to be really usefull! I will >>> try to learn some SQL code.. But actually I think I am really screwed... >>> >>> Because working with gds is not enough... I will nedd to work with gse >>> also, specially the ones that are no associated with a gds.. >>> I am working with mus musculus and I need around 1000 different >>> expression >>> tables. When I search for mus musculus in >>> https://www.ncbi.nlm.nih.gov/sites/GDSbrowser/ >>> it returns around 1300 DataSet records and I could only find around 250 >>> that have more than 20 samples... And then I have to filter the ones >>> that I >>> can identify possible transformations: log2, log10, etc >>> >>> I know the description is broad but that is what was asked for me to >>> do... >>> >>> So I am starting to belive I will have to work with data one by one in R >>> to >>> choose states :( >>> >>> >>> On Wed, Jul 24, 2013 at 4:43 PM, Tim Triche, Jr. <tim.triche@gmail.com>>> >wrote: >>> >>> > Incidentally (picking up where the previous example left off), one can >>> > catalog the number of subsetted GDS records and the states by which >>> they >>> > were subsetted as follows (assuming I have correctly understood your >>> > meaning for "states"): >>> > >>> > gds.with.subsets <- dbGetQuery(con, 'SELECT COUNT(DISTINCT gds) AS >>> > subsetted_datasets, type AS state FROM gds_subset GROUP BY type') >>> > >>> > gds.with.subsets >>> > ## data frame with 24 rows and 2 columns >>> > ## subsetted_datasets state >>> > ## <integer> <character> >>> > ## 1 184 age >>> > ## 2 866 agent >>> > ## 3 145 cell line >>> > ## 4 233 cell type >>> > ## 5 136 development stage >>> > ## ... ... ... >>> > ## 20 231 strain >>> > ## 21 112 stress >>> > ## 22 10 temperature >>> > ## 23 776 time >>> > ## 24 330 tissue >>> > >>> > Again, it may take a bit of SQL to assemble the results such that you >>> can >>> > troll through them with GEOquery, but I suspect that the process will >>> be >>> > less painful than if you did this manually through GEO2R. But that's >>> > something for you to decide. >>> > >>> > Best, >>> > >>> > --t >>> > >>> > >>> > >>> > On Wed, Jul 24, 2013 at 12:33 PM, Tim Triche, Jr. < >>> tim.triche@gmail.com>wrote: >>> > >>> >> You might want to look in gds_subset instead, for instances where >>> defined >>> >> differences exist. If you're searching free-form for arbitrary >>> differences >>> >> then yes, you're pretty much screwed. But for instances where GEO2R >>> can >>> >> see the differences, GEOmetadb can see them as well. You may need to >>> poke >>> >> around a bit, though, since your description is so broad (nothing >>> personal, >>> >> it's just an incredibly broad description and you may be the only >>> person >>> >> who can narrow it). >>> >> >>> >> I'm cc:'ing Sean on this, since he's more familiar with the package >>> (his >>> >> package!) than I am, but here's a start... >>> >> >>> >> >>> >> library(GEOmetadb) >>> >> geometadbfile <- getSQLiteFile() ## if needed >>> >> con <- dbConnect(SQLite(), geometadbfile) >>> >> >>> >> colDescs <- columnDescriptions() >>> >> colDescs[ grep('gds_subset', colDescs[,1]), ] ## ooh, this looks >>> >> promising... >>> >> >>> >> gds.res <- dbGetQuery(con, 'SELECT DISTINCT gds, type, description, >>> >> count(sample_id) AS samples FROM gds_subset GROUP BY type, gds') >>> >> >>> >> gds.res >>> >> ## data frame with 5498 rows and 4 columns >>> >> ## gds type description >>> samples >>> >> ## <character> <character> <character> >>> <integer> >>> >> ## 1 GDS1054 age 36 wk >>> 5 >>> >> ## 2 GDS1055 age 36 wk >>> 5 >>> >> ## 3 GDS1079 age 23 mo >>> 2 >>> >> ## 4 GDS1264 age 20 m >>> 3 >>> >> ## 5 GDS1265 age 22 d >>> 5 >>> >> ## ... ... ... ... >>> ... >>> >> ## 5494 GDS910 tissue bone marrow >>> 5 >>> >> ## 5495 GDS917 tissue whole blood >>> 2 >>> >> ## 5496 GDS953 tissue non-branching region >>> 2 >>> >> ## 5497 GDS956 tissue substantia nigra pars compacta >>> 2 >>> >> ## 5498 GDS960 tissue right lung >>> 2 >>> >> >>> >> >>> >> It may take some further twiddling on your part but I'd be surprised >>> if >>> >> this can't be pressed into service for your needs... It could take a >>> while >>> >> to extract 5498 subsets of data from GEO2R ;-) >>> >> >>> >> Best, >>> >> >>> >> >>> >> --t >>> >> >>> >> >>> >> >>> >> On Wed, Jul 24, 2013 at 11:58 AM, Lina Thomas <linadth6@gmail.com> >>> wrote: >>> >> >>> >>> Hello, thanks for your answer! >>> >>> >>> >>> I am studying the alterations of gene expression when a biological >>> >>> system suffers a change. So any change, any difference can be a >>> state. You >>> >>> can also call it a condition. >>> >>> >>> >>> With GDS I know where to find these information (disease.state or >>> >>> infection or stress or tissue, etc), but GSE is giving me a little >>> >>> headache. I am starting to think the information is hidden in the >>> fields >>> >>> title or source name (the two columns in GEO2R). But I can`t think >>> of an >>> >>> automatic way to separate the data... >>> >>> >>> >>> I took a quick look at GEOmetadb but it does not seem that is quite >>> what >>> >>> I need. It might be helpful in the future though. Thanks anyway! >>> >>> >>> >>> See ya, >>> >>> >>> >>> Lina >>> >>> >>> >>> >>> >>> On Wed, Jul 24, 2013 at 3:02 PM, Tim Triche, Jr. < >>> tim.triche@gmail.com>wrote: >>> >>> >>> >>>> how do you define "states"? >>> >>>> >>> >>>> You might find the GEOmetadb package very useful. Read the vignette >>> >>>> for it. It may be what you need. >>> >>>> >>> >>>> In terms of processing GSEs and GSMs, once you know what you are >>> >>>> searching for, it's much easier to find it. >>> >>>> >>> >>>> Hope this helps. >>> >>>> >>> >>>> --t >>> >>>> >>> >>>> >>> >>>> >>> >>>> On Wed, Jul 24, 2013 at 10:55 AM, Lina Thomas [guest] < >>> >>>> guest@bioconductor.org> wrote: >>> >>>> >>> >>>>> >>> >>>>> I am trying to work with GSE expression data from GEO in R using >>> >>>>> GEOquery. >>> >>>>> >>> >>>>> I am using the following commands >>> >>>>> gset <- getGEO("GSE52",GSEMatrix =TRUE,) >>> >>>>> ex <- exprs(gset[[1]]) >>> >>>>> >>> >>>>> But I have no idea how to separate the data in two different >>> states. I >>> >>>>> am trying to locate states data using >>> >>>>> >>> >>>>> varLabels(gset[[1]]@phenoData), when GSEMatrix =TRUE >>> >>>>> >>> >>>>> or >>> >>>>> >>> >>>>> Meta (gse), when GSEMatrix =FALSE >>> >>>>> >>> >>>>> But I still have no clue. I need to do this for around 1000 >>> different >>> >>>>> database, so I cannot do it using GEO2R manually... Is there a >>> particular >>> >>>>> field I can consider state? >>> >>>>> >>> >>>>> I also would like to know I can find out if data is already log2 >>> >>>>> transformed or log10 or inits raw form using just R. >>> >>>>> >>> >>>>> -- output of sessionInfo(): >>> >>>>> >>> >>>>> none >>> >>>>> >>> >>>>> -- >>> >>>>> Sent via the guest posting facility at bioconductor.org. >>> >>>>> >>> >>>>> _______________________________________________ >>> >>>>> Bioconductor mailing list >>> >>>>> Bioconductor@r-project.org >>> >>>>> https://stat.ethz.ch/mailman/listinfo/bioconductor >>> >>>>> Search the archives: >>> >>>>> http://news.gmane.org/gmane.science.biology.informatics.conductor >>> >>>>> >>> >>>> >>> >>>> >>> >>>> >>> >>>> -- >>> >>>> *A model is a lie that helps you see the truth.* >>> >>>> * >>> >>>> * >>> >>>> Howard Skipper< >>> http://cancerres.aacrjournals.org/content/31/9/1173.full.pdf> >>> >>>> >>> >>> >>> >>> >>> >> >>> >> >>> >> -- >>> >> *A model is a lie that helps you see the truth.* >>> >> * >>> >> * >>> >> Howard Skipper< >>> http://cancerres.aacrjournals.org/content/31/9/1173.full.pdf> >>> >> >>> > >>> > >>> > >>> > -- >>> > *A model is a lie that helps you see the truth.* >>> > * >>> > * >>> > Howard Skipper< >>> http://cancerres.aacrjournals.org/content/31/9/1173.full.pdf> >>> > >>> >>> [[alternative HTML version deleted]] >>> >>> _______________________________________________ >>> Bioconductor mailing list >>> Bioconductor@r-project.org >>> https://stat.ethz.ch/mailman/listinfo/bioconductor >>> Search the archives: >>> http://news.gmane.org/gmane.science.biology.informatics.conductor >>> >> >> > > > -- > *A model is a lie that helps you see the truth.* > * > * > Howard Skipper<http: cancerres.aacrjournals.org="" content="" 31="" 9="" 1173.full.pdf=""> > [[alternative HTML version deleted]]