How did you determine the threshold "0.335"

jaswindersingh2 / SPOT-RNA

RNA Secondary Structure Prediction using an Ensemble of Two-dimensional Deep Neural Networks and Transfer Learning.

https://apisz.sparks-lab.org:8443/spot-rna-sz.html

Mozilla Public License 2.0

94 stars 33 forks source link

How did you determine the threshold "0.335" #13

Open HarveyYan opened 3 years ago

HarveyYan commented 3 years ago

Kudos for the nice work. One lttle question I have wondered whilst reading your paper was about that threshold which converts probabilities to basepairs. Would you mind clarifying a bit how it was chosen?

jaswindersingh2 commented 3 years ago

Hi Yan,

The threshold was determined by maximizing the Matthews Correlation Coefficient (MCC) on validation set (VL1). MCC was calculated for 1000 values between 0 and 1 at sampling rate 0.001. MCC on validation set was maximum at 0.335, therefore, we selected this value. All the model parameters were optimized for the validation sets.

Jaswinder

HarveyYan commented 3 years ago

Thanks for the prompt reply!