The catRAPID method

Our approach

In our method, proteins and RNAs are associated with wave-packets that contain information on their binding ability.

The wave-packets are built using phenomenological potentials for secondary structure, hydrogen bonding and van der Waals propensities.

The interference between protein and RNA waves determines the binding probability of molecules (i.e., the catRAPID score).

In analogy with the Standard Model of Particle Physics, proteins and RNAs can be regarded as particles carrying a signal.

In this view, waves are associated with a "riboproteon" signal (analogous to the force carrier "boson").


About the algorithm

We trained our method using a positive and a negative set of protein-RNA associations.

We generated the interaction matrix by requiring that the combination of protein and RNA signals is maximal for the positive set. In the optimization phase, the negative set is merged with the positive set and the algorithm learns to associate maximal scores to the positive set.

It should be noted that the interaction matrix is obtained in Fourier’s transform space to reduce the number of coefficients to be derived.