Title:  How entangled proteins can be? Prediction and in vitro verification.
Abstract:  The presence of knots in protein 3D structures is well established [1]. However, as they are not common [2], only a fraction of such proteins is available in the PDB database. It was not possible to assess their im portance and versatility up until now because we did not have access to the whole proteome of an organ ism. The development of ecient machine learning methods for protein structure prediction, such as Al phaFold and RoseTTaFold, changed that [3].

First, we analyzed all proteins from human proteome (over 20 000) in search for knots and found them in less than 2% of the structures. Using a variety of methods, including homolog search, clustering, quality assessment and visual inspection, we determined the nature of each of the knotted structures and clas sifed it as either knotted, potentially knotted or an artifact and de-posited in a database available at: https://knotprot.cent.uw.edu.pl/alphafold. It turns out that over 75% of knots in human proteins may be artifacts. However, among potential knots we found knot 63 which would be the first knot in proteins with unknotting number higher than one [4].

Second, we focused on proteins which can be even more complex and possess two knots on a single chain (up to now, only single knots were found). For the first time, we searched dierent databases for double knotted proteins. Using AlphaFold we predicted a few families of doubly knotted proteins and studied in de tail their structure and function. Using experimental approach, we showed that such proteins can fold and perform their intended function 3, Fig. 1.