Table 6.

Model evaluation

Require: A thermodynamic model T, an input set S with perfect match DNA duplexes.

Ensure: Vectors of Pearson correlations () and root mean square errors () for all duplexes.

1: Initialize correlations vector

2: Initialize root mean square errors vector

3: for i = 0 to 10 000 do

4:    Training set TrS = 67% of randomly chosen data from S

5:    Testing set TeS = remaining 33% of data from S

6:    Train model T on data in TrS

7:    Compute r and RMSE for each data point in TeS

8:

9:

10: end for

11: return vectors and