Hi Tam, Sorry about the confusion. Two items: 1. Adding a column of numbers (row number) is a a known "harmless" bug in the ComBat function. It is basically adding a row of column numbers to your dataset, which can easily just be deleted in Excel (and the data can be shifted over). However I recognize that your case is a little different, hence: 2. On line 3154, your gene description has some character formatting that is causing issues with R's "read.table" function--which is used by ComBat. I think its reading the apostrophe as quoted text, so it is then concatenating everything after that as text until the next apostrophe. Anyway, here is how you fix it: open up your dataset ('12arraysCombatImputed_2.txt') in Excel, then save it as a .csv. Then run ComBat using the option: type='csv'. Alternatively, you can remove the second column from your dataset for ComBat and then add them back in after adjustment. I just did this myself on your data and it worked. However, you still need to delete the column numbers from item #1 before the data are ready to go! Also, I noticed that your variances are also not well-behaved (see the plot that comes up), so I'd recommend that you use a non-parametric prior (par.prior=F). Note that this may take an hour or so to run so make sure that the parametric prior is working before you try the non- parametric one. Thanks! Evan Okay, I looked at your data. On line 3154, R's "read.table" On Oct 31, 2012, at 10:33 AM, SSc Array Core wrote: > I am running Combat on the attached files. 2 channel array (with reference). Problem is the adjusted file returns an added column of numbers where CLID should be. This column then stops delivering said numbers around line 3154, returning to CLID, shifting all the information and data to the left. I am stumped as to why this is happening. Please advise. > > thanks, > > tam > > Reading Sample Information File > Reading Expression Data File > Found 2 batches > Found 1 covariate(s) > Found 260 Missing Data Values > Standardizing Data across genes > Fitting L/S model and finding priors > Finding parametric adjustments > Adjusting the Data > Adjusted data saved in file: Adjusted_12arraysCombatImputed_2.txt_.xls > > ComBat('12arraysCombatImputed_2.txt','sample_info_file_mouse.txt', skip=2,write=T) > <12arraysCombatImputed_2.txt><sample_info_file_mouse.txt><adjusted_1 2arrayscombatimputed_2.txt_.xls="">

Hi, > 2. On line 3154, your gene description has some character formatting > that is causing issues with R's "read.table" function--which is used > by ComBat. I think its reading the apostrophe as quoted text, so it > is then concatenating everything after that as text until the next > apostrophe. > Anyway, here is how you fix it: open up your dataset > ('12arraysCombatImputed_2.txt') in Excel, then save it as a .csv. > Then run ComBat using the option: type='csv'. > Alternatively, you can remove the second column from your > dataset for ComBat and then add them back in after adjustment. To make life easier for ComBat users and to avoid similar problems in the future, quote="" should be added to the read.table() call in ComBat. Regards Gerhard ---------------------------------------------------------------------- -- Dr. Gerhard Thallinger E-mail: Gerhard.Thallinger at tugraz.at Institute for Genomics and Bioinformatics Web: http://genome.tugraz.at Graz University of Technology Tel: +43 316 873 5343 Petersgasse 14/V Fax: +43 316 873 105343 8010 Graz, Austria Map: http://genome.tugraz.at/Loc.html

Hey Simona, Thanks for your email. This is a difficult question. If the number of samples and number of miRNAs in the data are large, then I think it would be safe to apply ComBat to the data. The non-parametrix prior option might be best. For small sample or miRNA sizes, I don't think there is a good tool out there at the moment. I'd be interested to see how it works on your data. Thanks! Evan On Nov 12, 2012, at 3:45 PM, Simona Rossi wrote: > Hello, > > I'm Simona and I'm a Bioinformatician. > > I'm studying some miRNASeq data that are affected by a > strong batch effect, in your opinion, ComBat might be > applied to miRNASeq data? > > Thank you very much in advance, > > Best Regards, > > Simona Rossi > > -- > -- > Dr. Simona Rossi > Bioinformatician > > SIB | Swiss Institute of Bioinformatics > > Quartier Sorge > Bâtiment Génopode > CH-1015 Lausanne > Switzerland > > simona.rossi@isb-sib.ch > www.isb-sib.ch > [[alternative HTML version deleted]]