Skip to main content

Table 1 Comparison of the accuracy of the Janssens model and our model.

From: Studying the functional conservation of cis-regulatory modules and their transcriptional output

  Cross validation
Model Training error (SD) Testing error (SD)
Janssens model 8.89 (0.34) 12.95 (2.33)
Model using predicted TFBS 9.38 (0.35) 13.06 (2.41)
  1. The table shows the average (standard deviation) RMS error for three independent training runs using the Janssens model (knowledge-based TFBS map and extra free parameters) and our model (predicted-only TFBS map and no extra parameters), respectively. Training was done using the simulated annealing algorithm. For the model using predicted sites, we used a log-odds threshold of 9 bits and the D. melanogaster background base frequencies with a total pseudo-count of 1.5.