Dear Mousami, Thanks for your email and interest in the SpeCond package. You don't need to worry about the above warnings. These warnings are returned by the mclust package that the SpeCond method uses to fit a mixture of normal distribution. I restricted the number of normal (or component) to be tested from 1 to 3. By default if the best model is found at the min or max number of the component tested the mclust function will return a warning message. However for the SpeCond method testing only 1 to 3 components makes sense, according to the pattern the method is looking for to identify specifically expressed gene/probes. I should probably add in the vignette that one can ignore these warnings (which also appear when running the test example). I would advice to plot some HTML pages for a few genes (or probes) to check if the mixture of normal components does fit your data (and play with the parameters if not). The error returned by "getMatrixFromExpressionSet", probably comes from the fact that condition.factor=eset.mas5$sample is not of class factor (the factor type is required by the function). You can checkt it by running class(eset.mas5$sample). Running SpeCond on a ExpressionSet object makes sense if you have several samples (replicates) for the same condition (or tissue for example) and you have to summarize the gene or probe values per condition but if you only have one sample per condition (which seems to be your case) you can just extract the expression matrix from the ExpressionSet object using the exprs() (which is what you probably have done before). In the vignette example there are two replicates per tissue as you can see below, that is why it is necessary to run getMatrixFromExpressionSet to get a matrix with one value per probe per tissue type. > class(expSetSpeCondExample$Tissue) [1] "factor" > expSetSpeCondExample$Tissue S_1 S_2 S_3 S_4 S_5 Spinal_cord Spinal_cord Fetal_brain Fetal_brain Adrenal_cortex S_6 S_7 S_8 S_9 S_10 Adrenal_cortex Pituitary Pituitary Whole_brain Whole_brain S_11 S_12 S_13 S_14 S_15 Lymph_node Lymph_node Liver Liver Prostate S_16 S_17 S_18 S_19 S_20 Prostate Uterus Uterus Fetal_lung Fetal_lung S_21 S_22 S_23 S_24 S_25 Appendix Appendix Fetal_thyroid Fetal_thyroid Heart S_26 S_27 S_28 S_29 S_30 Heart Lung Lung Ovary Ovary S_31 S_32 S_33 S_34 S_35 Placenta Placenta Pancreas Pancreas Skeletal_muscle S_36 S_37 S_38 S_39 S_40 Skeletal_muscle Smooth_muscle Smooth_muscle Tongue Tongue S_41 S_42 S_43 S_44 S_45 Thyroid Thyroid Tonsil Tonsil Whole_blood S_46 S_47 S_48 S_49 S_50 Whole_blood Adrenal_gland Adrenal_gland Bone_marrow Bone_marrow S_51 S_52 S_53 S_54 S_55 Fetal_liver Fetal_liver Kidney Kidney Salivary_gland S_56 S_57 S_58 S_59 S_60 Salivary_gland Skin Skin Testis Testis S_61 S_62 S_63 S_64 Thymus Thymus Trachea Trachea 32 Levels: Adrenal_cortex Adrenal_gland Appendix Bone_marrow ... Whole_brain Hope this helps. Please let me know if you have further questions, Best wishes, Florence 2013/11/26 Mousami Srivastava <sm.iitbtbi@gmail.com> > Dear Florence, > > Greetings from Mousami, working as a Ph.D Student in Defence Institute of > Physiology and Allied Sciences, India. > > I have found that your program "specond" most suitable for analysing my > data of interest to find their shannon entropy. During this analysis, i am > getting the following warning message even after many repeated attempts, > please help/advice me (how and where) to rectify the error > > Warning messages while running the log transformed file > > 1: In summary.mclustBIC(Bic, data, G = G, modelNames = modelNames) : > best model occurs at the min or max # of components considered > 2: In Mclust(expressionMatrix[i, ], G = 1:3, modelNames = "V") : > optimal number of clusters occurs at min choice > > Warning messages while running the normalized file > > In case of "condition.factor=expSet$sample", I have got the following > error: > "Error in getMatrixFromExpressionSet(expressionMatrix, condition.factor, > : > condition must be of class factor" > > The code of the program is: > > class(eset.mas5) > > [1] "ExpressionSet" > attr(,"package") > [1] "Biobase" > > > eset.mas5 > > ExpressionSet (storageMode: lockedEnvironment) > assayData: 22283 features, 23 samples > element names: exprs, se.exprs > protocolData > sampleNames: GSM286407.CEL.gz GSM286408.CEL.gz ... GSM286429.CEL.gz > (23 total) > varLabels: ScanDate > varMetadata: labelDescription > phenoData > sampleNames: GSM286407.CEL.gz GSM286408.CEL.gz ... GSM286429.CEL.gz > (23 total) > varLabels: sample > varMetadata: labelDescription > featureData: none > experimentData: use 'experimentData(object)' > Annotation: hgu133a > > > class(exprSet_mat) > > [1] "matrix" > > > dim(exprSet_mat) > > [1] 22283 23 > > > library(SpeCond) > > > generalResult=SpeCond(exprSet_mat, param.detection=NULL, > multitest.correction.method="BY", > + prefix.file="E", print.hist.pv=TRUE, fit1=NULL, fit2=NULL, > specificOutlierStep1=NULL) > > [1] "Step1" > [1] "Step1, fitting" > [1] "start: get null distributions" > [1] "end: get null distributions" > [1] "start: specific detection from p-values" > [1] "end: specific detection from p-values" > [1] "Step2" > [1] "Step2, fitting" > [1] "start: get null distributions" > [1] "end: get null distributions" > [1] "start: specific detection from p-values" > [1] "end: specific detection from p-values" > There were 50 or more warnings (use warnings() to see the first 50) > > > warnings() > > Warning messages: > 1: In summary.mclustBIC(Bic, data, G = G, modelNames = modelNames) : > best model occurs at the min or max # of components considered > 2: In Mclust(expressionMatrix[i, ], G = 1:3, modelNames = "V") : > optimal number of clusters occurs at min choice > 3: In summary.mclustBIC(Bic, data, G = G, modelNames = modelNames) : > best model occurs at the min or max # of components considered > > > names(generalResult) > [1] "prefix.file" "fit1" "fit2" > "specificOutliersStep1" "specificResult" > > specificResult=generalResult$specificResult > > generalResult=SpeCond(eset.mas5, param.detection=NULL, > multitest.correction.method="BY", prefix.file="E", print.hist.pv=TRUE, > fit1=NULL, > + fit2=NULL, specificOutlierStep1=NULL, condition.factor=eset.mas5$sample, > condition.method="mean") > Error in getMatrixFromExpressionSet(expressionMatrix, condition.factor, : > condition must be of class factor > > > > > Many thanks > -- > Regards, > Mousami Srivastava > Email : sm.iitbtbi@gmail.com > srvstvmsm@yahoo.com > Mob. No.: +91-9555896841 > Research Scholar > DIPAS, DRDO > Delhi > India > [[alternative HTML version deleted]]