Bioinformatics prediction of 3D protein structure

Predicting the 3D structure of a protein from its sequence alone is an ambitious challenge. Even if secondary structures can be predicted with fairly high reliability, predicting the relative positioning of these structures in 3D remains very difficult. Take the example of a protein predicted to consist of three helices and three strands, and the loop connections between these elements: the diversity of pairing of these elements is extremely vast, because under the terminology "loop", conformations are extremely numerous. Even if the enumeration of these possibilities can be considered exhaustive, sorting through them and identifying the one corresponding to the native structure requires a function capable of precisely quantifying the relevance of this structure, in other words a score function. This has only been attempted for small sequences (a few dozen amino acids). For proteins, it...