You're probably right. Look at figure 3 in the paper. The effect is for region upstream of FOXP4. The reported Prometheus snp ( rs1983891 ) is not in this region, it is in an intron in FOXP4.