I think GLIMMIX can not include different families in the same one model? Am I wrong? ... View more

Hello @SteveDenham  your descriptions of my study design are completely correct. I think my experiment design is very simple. Two treatments (1 control and 1 antibiotics), and animals are randomly assigned into two treatments according to RCBD. Now both cecal and fecal samples are collected and are analyzed for microbiota under kingdom, phylum, class, order, family, genus, and species levels. The family, genus, and species levels have so many bacteria that I have to compare them one by one between two treatments. That is the tricky thing.   The percentage data I have converted through excel and can be uploaded to SAS. I here also include my data input code for your reference. data cecal_family_taxonomy; length treatment $20; Infile "/home/u39233094/sasuser.v94/Thesis/CSV file/5-family cecal taxonomy for SAS.csv" dlm="," firstobs=5; input diet$ treatment$ block pen pig_number Family1 Family1P Family2 Family2P Family3 Family3P Family4 Family4P....Family121P Family122 Family122P TotalR TotalP; /*species have over 800 bacteria that have to be compared one by one*/ run; ... View more

My analyses are about gut microbiome. Even though there are large amount of outputs, I have to look through them one by one to see how my positive treatment has an impact on specific bacteria that is under family or species level compared to control group. Even there are over 800 family, I might only find let is say 20 something that have significant differences. Don't know yet.  ... View more

Hello @PaigeMiller, the codes I attached at the beginning include the tests I need to run. Basically I need to have homogeneity test results, then I do an Lsmean comparison (F-test) between my two treatment groups. But now I have many families that I need to repeat the same. I know there is now BY or macro statement I can choose, but so far, I don't know how to write into my codes to be specific.  proc glimmix data= cecal_family_taxonomy; class treatment block; model Family1P = treatment; /*here I have to repeat family2P, 3P...until 800P*/ random block; random _residual_/group=treatment; lsmeans treatment/tdiff lines; covtest homogeneity; ods output LSMeans = myparmdataset; output out=second predicted=pred residual=resid residual(noblup)=mresid student=sresid student(noblup)=smresid; title "1-Taxonomy Data Cecal Family 1 Percentage ANOVA Results"; run; proc print data=myparmdataset; format estimate D8.6 stderr D8.6; run; proc print data=second; run; proc sgplot data=second; scatter y=smresid x=pred; refline 0; run; proc sgplot data=second; scatter y=smresid x=treatment; refline 0; run; proc sgplot data=second; vbox smresid/group=treatment datalabel; run; proc sgscatter data=second; plot sresid*(pred treatment block); run; proc univariate data=second normal plot; var sresid; histogram sresid / normal kernel; run; ... View more

Hello @PaigeMiller. Yes, there are many good advice and I really appreciate it!  But I still have difficulty in adapting those advice into my own codes. Need some time to digest them. Thank you very much again for your kind help.  ... View more

Hello @SteveDenham You are right, my data are presented as percentages (counts of family X/total count). But I don't have any experience with the BY statement and macro statement. If possible, do you mind giving me an example of how to insert relevant codes into my procedure? My supervisor always wants to see my homogeneity test results before the comparisons of treatments. That's why I keep this several lines of codes here.      ... View more

@PaigeMiller  It is several hundred variables in one data set.  ... View more