Hello, Martin! I am looking for the spot in the algorithm that is the reason for the asymmetry in blast. I'm not sure if I got your argument but > > Where does the background distribution of the amino acid (or in my case DNA) > > sequence of the query come in? > > Spw_ij = log( Q_ij / (P(i) P(j) )) / λ this is used for the construction of a scoring matrix. The frequencies here are not taken from the actual blast search. This is the formula with which lambda is estimated: > For λ the equation > > sum_i,j P(i) P(j) exp(λSij) = 1 > > must hold. > This would give rise to asymmetry if P(i) and P(j) are only calculated from the database sequence and the query is ignored. Is this the case? Regards, Michael.