Table 6

Log-likelihoods of alignments, and log-posteriors of alignment annotations, for training and testing datasets using the original Pfold program. Comparison with Table 5 shows that EM training increases all probabilities, as desired.

Dataset

log₂ P(D, A|θ)

log₂ P(D|θ)

log₂ P(A|D, θ)

Training set

-487422.5964

-464828.9148

-22593.6816

Test set

-370490.5284

-348550.7516

-21939.7768

Klosterman et al. BMC Bioinformatics 2006 7:428   doi:10.1186/1471-2105-7-428