Slam predicts coding exons and conserved noncoding regions in a pair of homologous DNA sequences, incorporating both statistical sequence properties and degree of conservation into predictions. The model is symmetric and the same structure (with possibly different lengths) is predicted in both sequences.
The CNS (conserved non-coding sequence) predictions are ab-initio predictions of conserved regions that do not fit in with a gene structure. Thus, slam is not simply trying to predict conserved regions to be coding, but is classifying such regions according to an overall probabilistic model of gene structure. The set of slam CNS predictions is therefore highly enriched for conserved non-coding regions.
More information can be found at http://math.berkeley.edu/~lpachter/software.html.
M. Alexandersson, S. Cawley, L. Pachter (2003). SLAM - Cross-species Gene Finding and Alignment with a Generalized Pair Hidden Markov Model. Genome Research 13(3):496-502.
L. Pachter, M. Alexandersson, S. Cawley (2001). Applications of Generalized Pair Hidden Markov Models to Alignment and Gene Finding Problems, Proceedings of the Fifth Annual International Conference on Computational Molecular Biology (RECOMB 2001).
L. Pachter , M. Alexandersson, S. Cawley (2002). Applications of Generalized Pair Hidden Markov Models to Alignment and Gene Finding Problems, Journal of Computational Biology 9(2):389-400.