New Deep Learning Methods for Protein Loop Modeling

Abstract

Computational protein structure prediction is a long-standing challenge in bioinformatics. In the process of predicting protein 3D structures, it is common that parts of an experimental structure are missing or parts of a predicted structure need to be remodeled. The process of predicting local protein structures of particular regions is called loop modeling. In this paper, five new loop modeling methods based on machine learning techniques, called NearLooper, ConLooper, ResLooper, HyLooper1 and HyLooper2 are proposed. NearLooper is based on the nearest neighbor technique; ConLooper applies deep convolutional neural networks to predict Cα atoms distance matrix as an orientation-independent representation of protein structure; ResLooper uses residual neural networks instead of deep convolutional neural networks; HyLooper1 combines the results of NearLooper and ConLooper while HyLooper2 combines NearLooper and ResLooper. Three commonly used benchmarks for loop modeling are used to compare the performance between these methods and existing state-of-the-art methods. The experiment results show promising performance in which our best method improves existing state-of-the-art methods by 28% and 54% of average RMSD on two datasets while being comparable on the other one.

Structural dissection of sequence recognition and catalytic mechanism of human LINE-1 endonuclease

“…First, the chimeric model was built by ‘morphing’ the Mg 2+ bound structure outside the immediate vicinity of the Mg 2+ ion (12 Å) into the DNA-bound conformation. To accomplish this, harmonic tethers to the DNA-bound structure were applied and constrained force-field relaxation (ICMFF forcefield ( 62 )) in internal coordinates was performed, while the active site region was constrained to the Mg 2+ bound structure coordinates. Next, the resulting enzyme structure was combined with DNA in three alternative conformations: (i) the B-form conformation observed in all complexes, (ii) the alternative rotated conformation of DNA observed in the complex with dna14 or (iii) an alternative model of the polynucleotide chain containing the -O-[PO 3 H] –2 -O- (holophosphoric acid diester) link in the transition state based on the alternative rotated conformation.…”

Deep Learning in Protein Structural Modeling and Design

“…Loops are irregular and sometimes flexible segments, and thus their structures have been difficult to capture experimentally or computationally. 176 , 177 So far, DL frameworks based on inter-residue distance prediction (similar to protein structure prediction), 178 and those based on treating the loop residue distances with the remaining residues as an image inpainting problem 179 have been applied to loop modeling. Recently, Ruffolo et al.…”

Aplicaciones de las redes neuronales y el deep learning a la ingeniería biomédica

“…La predicción de la estructura de una proteína es crucial para determinar su función y aplicación, por lo tanto, dentro de la proteómica y la epigenómica, se ha investigado en aplicaciones que permiten la predicción de los ángulos de la columna vertebral de la proteína (que permite obtener información sobre la estructura de la proteína) [36][37][38], la predicción de la estructura secundaria y terciaria de la proteína [39][40][41][42][43][44][45][46][47], la evaluación de la calidad de la proteína [48][49][50], y la predicción de las regiones de desórdenes (trastornos) en la estructura de la proteína [51][52].…”