/*normaliz macro is designed to calculate several different        */
/*normalization schemes for the signal intensities of              */
/*microarray analyses.                                             */
/*  The following normalizations are performed:                    */
/*	natural log of the signal intensity                        */
/*	log base 10 of the signal intensity                        */
/*	divided by the median within the specified class(es)       */
/*	natural log and divided by the median normalized           */
/*	square root of the signal intensity                        */
/*	rank normalized                                            */
/*  standardized                                                   */
/*	quartile normalized                                        */

%macro normaliz(infile,signal,factor1,factor2,factor3,suffix);
	data add_log;
	set &infile;
		log_&signal=log(&signal);
	run;

	proc means data=&infile median noprint;
		class &factor1 &factor2;
		var &signal log_&signal;
		output out=&factor1&factor2&factor3.med 
			median(&signal)=median q1(&signal)=q1 
			q3(&signal)=q3 min(&signal)=min 
			max(&signal)=max p95(&signal)=p95 
			p99(&signal)=p99 mean(log_&signal)=mean_log
			std(log_&signal)=std_log;
	run;

	data &factor1&factor2&factor3.med1;
	set &factor1&factor2&factor3.med;
		where _type_=3;
	run;
 
	proc sort data=&infile;
		by &factor1 &factor2 &factor3;
	run;

	data med_merge;
		merge &factor1&factor2&factor3.med1 &infile;
		by &factor1 &factor2 &factor3;
	run;

	data quartiles;
	set med_merge;
		if &signal < q1 then &signal._quartile=1;
		if &signal ge q1 and &signal < median
			then &signal._quartile=2;
		if &signal ge median and &signal < q3
			then &signal._quartile=3;
		if &signal ge q3 then &signal._quartile=4;
		sqrt_&signal=sqrt(&signal);
		log10_&signal=log10(&signal);
		&signal._med&suffix=&signal/median;
		log_&signal._med&suffix=log(&signal._med&suffix);
		std_&signal=(log_&signal-mean_log)/std_log;
		drop _type_ _freq_ mean_log std_log;
	run;

	proc rank data=quartiles out=anova_&signal._norm;
		by &factor1 &factor2 &factor3;
		var &signal;
		ranks &signal._rank;
	run;
%mend normaliz;