stephenturner · March 31, 2011 03:22
diff --git a/2011-03-30 igf1 stufy maf vs 1000g maf.r b/2011-03-30 igf1 stufy maf vs 1000g maf.r
 d=query("-- DISCOVERED IN STUDY, BUT SOME MIGHT NOT BE IN 1000 GENOMES
 select distinct *, CASE WHEN refvar_study!=refvar_1000g and maf1000g IS NOT NULL THEN 1 ELSE 0 END as mismatch from
 (SELECT a.chr, a.pos18, c.pos19, a.major||a.minor AS refvar_study, a.maxpoolmaf, c.major||c.minor AS refvar_1000g, c.maf as maf1000g
 FROM igf1_pooledmaf a 
 LEFT JOIN igf1_map1819 b ON a.pos18=b.hg18 
 LEFT JOIN igf1_1000g_freq c ON b.hg19=c.pos19);")

 nrow(d)
 ht(d)

 require(ggplot2)
 p=qplot(maf1000g,maxpoolmaf, data=d, xlab="1000 Genomes MAF", ylab="Study MAF", main="Study MAF vs 1000G MAF") + geom_smooth(method="lm")
 ggsave("2011-03-30 study maf vs 1000G maf.png", p, w=6, h=6, dpi=100)

 #what's that odd point down there?
 subset(d,maf1000g>.5&maxpoolmaf<.2)

 d[d$mismatch==1,]
	d=query("-- DISCOVERED IN STUDY, BUT SOME MIGHT NOT BE IN 1000 GENOMES
	select distinct *, CASE WHEN refvar_study!=refvar_1000g and maf1000g IS NOT NULL THEN 1 ELSE 0 END as mismatch from
	(SELECT a.chr, a.pos18, c.pos19, a.major\|\|a.minor AS refvar_study, a.maxpoolmaf, c.major\|\|c.minor AS refvar_1000g, c.maf as maf1000g
	FROM igf1_pooledmaf a
	LEFT JOIN igf1_map1819 b ON a.pos18=b.hg18
	LEFT JOIN igf1_1000g_freq c ON b.hg19=c.pos19);")

	nrow(d)
	ht(d)

	require(ggplot2)
	p=qplot(maf1000g,maxpoolmaf, data=d, xlab="1000 Genomes MAF", ylab="Study MAF", main="Study MAF vs 1000G MAF") + geom_smooth(method="lm")
	ggsave("2011-03-30 study maf vs 1000G maf.png", p, w=6, h=6, dpi=100)

	#what's that odd point down there?
	subset(d,maf1000g>.5&maxpoolmaf<.2)

	d[d$mismatch==1,]