I paid for the study done in 2005 on Somali men in Denmark, organized the STR markers to the ysearch standard, and I ran the numbers against the haplogroup predictor.
I got very different results than those of people who published in 2006 and 2007.
Fully 80% or 161 of the 201 people tested are E1b1b, 3 people tested for E1b1a (the "Bantu" haplogroup) 5 people tested for J1 (the "Arab" haplogroup), 19 people tested for Haplogroup T (the "Southern Iran" haplogroup) and there was 1 G2a (Circassian, Mameluke from Egypt maybe?), 1 J2b (Arab from Syria?), 1 I1 (Englishman?) 2 I2a (Italian?) 1 R1a (could be Sardinian or Indian) 1 H (Indian)
http://spreadsheets.google.com/pub?key= ... utput=html
The results indicate, at least to me, that most of these people are probably from Mogadishu. The very high number of T haplotypes would indicate somewhere with a significant minority of Persian-descended people.
As such, I do not believe that you can say Somalis are "x percent this, X percent that" because this survey tells us little.