I am using EMBOSS implementation of needleman to build a similarity matrix of 50k nucleotide reads against each other;

1. I don't know how to get the final alignment result from EMBOSS implementation of needleman (it returns the score) and I don't see any parameter to specify the alignment file in the output
2. What would be the most efficient way of doing this. 1 read against n-1 reads, for all n reads? (it doesn't sound so efficient, specially since the needleall, can handle alignment of many reads against many reads) ...