Hi,
I am conduction series of data analysis on faecal egg count (as indicator for resistant/susceptible animals) and its association with genotypes . We are using microsatellite markers (multi - allelic) for genotyping. I am able to find the effect of particular genotype on FEC using SPSS GLM , however, I want to find effect of particular allele in FEC and other indicator traits (continuous variable as well as binomial), I am having access to SPSS and SAS university edition only. How can i do it ?
PS - I am not familiar with much statistical softwares, only SPSS, SAS, details on how to prepare the data for allele wise analysis will be highly useful for me :-)
Please check it up:
Article Q&A: Genetic Analysis of Quantitative Traits
Good Luck.
First: I highly recommend that you read the Q&A in the link Mosayeb Dashtpeyma provided.
Second: Because quantitative gene expression usually relies on many loci and the relationship between those alleles at those loci are rarely additive, you need more information. What environmental factors influence FEC expression? Do some combination of alleles suppress expression while others promote? Presence or absence data may not be enough information either... Perhaps you may need to know the relative levels of expression for the underlying alleles.
If you can assume that the alleles underlying your quantitative trait (FEC?) have an additive effect on FEC, then you may consider each marker as a factor and build a linear model. If you have a sufficiently large sample, you may be able to model the interactions between alleles as well.
Finally: I know nothing about FEC, so I cannot recommend this procedure. In general, cases in which such linear models can be built are rare. QTL experiments are the best solution.
This is straight-forward for SNP data, as you can focus on one allele (and regress on the number of copies of that allele). For microsatellite data, you need to do the same thing for each allele, i.e. set up a variable that contains the number of copies of that allele, and regress on all those variables. I've included a SAS macro below, but it is some time since I used this. You will need to modify for your own use, but could get you started.
%MACRO assoc(marker);
%LET traits= trait1;
%LET fameffect= family ;
DATA allnames;
SET mydata;
LENGTH allele $1;
KEEP allele;
IF INDEXC(&marker,' /')=0 THEN DO;
allele=SUBSTR(&marker,1,1); OUTPUT;
allele=SUBSTR(&marker,2,1); OUTPUT;
END;
PROC SORT DATA=allnames FORCE NODUPKEY; BY allele;
PROC SQL NOPRINT; SELECT COUNT(*) INTO :nalls FROM allnames;
%LET nalls=&nalls;
DATA allnames;
SET allnames;
allnum=_N_;
KEEP allnum allele;
RETAIN allalls;
LENGTH allalls $&nalls;
allalls=COMPRESS(allalls)||COMPRESS(allele);
CALL SYMPUT('allalls',allalls);
DATA scores;
SET mydata;
DROP i;
ARRAY allscore{*} allsc1-allsc&nalls;
IF INDEXC(&marker,' /')^=0 THEN &marker=' ';
DO i=1 TO &nalls; allscore{i}=0; END;
i=INDEX("&allalls",SUBSTR(&marker,1,1)); IF i>0 THEN allscore{i}=allscore{i}+1;
i=INDEX("&allalls",SUBSTR(&marker,2,1)); IF i>0 THEN allscore{i}=allscore{i}+1;
%PUT Alleles for &marker : &allalls;
PROC GLM DATA=scores OUTSTAT=glmo1 NOPRINT;
CLASS &fameffect ▮
MODEL &traits = &fameffect allsc1-allsc&nalls &marker /SS1;
PROC SORT DATA=glmo1; BY _NAME_ _TYPE_;
PROC MEANS NOPRINT DATA=glmo1;
WHERE _SOURCE_^='ERROR';
BY _NAME_ _TYPE_;
VAR df ss;
OUTPUT OUT=glmo2 (DROP=_FREQ_ _TYPE_) SUM=;
PROC MEANS NOPRINT DATA=glmo1;
WHERE _SOURCE_=:'allsc';
BY _NAME_ _TYPE_;
VAR df ss;
OUTPUT OUT=glmo3 (DROP=_FREQ_ _TYPE_) SUM=;
DATA glmo3;
SET glmo2(IN=in2) glmo3 (IN=in3) glmo1(WHERE=(_SOURCE_="&marker"));
IF in2 THEN _SOURCE_='Genotype';
IF in3 THEN _SOURCE_='Additive';
PROC SORT; BY _NAME_;
DATA glmoe;
SET glmo1;
WHERE _TYPE_='ERROR';
RENAME df=dfe ss=sse;
KEEP df ss _NAME_;
DATA glmo3;
MERGE glmo3 glmoe;
BY _NAME_;
KEEP marker _NAME_ _SOURCE_ df ss f prob sse dfe;
marker="&marker";
IF _SOURCE_=marker THEN _SOURCE_='Non-Add';
f=(ss/df)/(sse/dfe);
prob=1-PROBF(f,df,dfe);
PROC PRINT DATA=glmo3 NOOBS;
RUN;
%MEND;
%assoc(Marker1);
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software