A fully open-source framework for deep learning protein real-valued distances

Abstract

As deep learning algorithms drive the progress in protein structure prediction, a lot remains to be studied at this emerging crossway of deep learning and protein structure prediction. Recent findings show that inter-residue distance prediction, a more granular version of the well-known contact prediction problem, is a key to predict accurate models. We believe that deep learning methods that predict these distances are still at infancy. To advance these methods and develop other novel methods, we need a small and representative dataset packaged for fast development and testing. In this work, we introduce Protein Distance Net (PDNET), a dataset derived from the widely used DeepCov dataset and consists of 3456 representative protein chains for training and validation. It is packaged with all the scripts that were used to curate the dataset, generate the input features and distance maps, and scripts with deep learning models to train, validate and test. Deep learning models can also be trained and tested in a web browser using free platforms such as Google Colab. We discuss how this dataset can be used to predict contacts, distance intervals, and real-valued distances (inÅ) by designing regression models. All scripts, training data, deep learning code for training, validation, and testing, and Python notebooks are available at https://github.com/ba-lab/pdnet/.

Protein sequence‐to‐structure learning: Is this the end(‐to‐end revolution)?

“…While most protein structure prediction methods take pre-computed features as input and output a contact or distance map, possibly augmented with other geometrical features (Fig. 1, see iPhord, ProSPr [34], Kiharalab_Contact [35], Pharmulator, DeepPotential, RaptorX [36], Galaxy, Triple-tRes [37], A2I2Prot, DESTINI2 [38], DeepHelicon [39], DeepHomo [40], ICOS, PrayogRealDistance [41,42], RBO-PSP-CP [43], DeepECA, ropius0 [44], tFOLD, plus QUARK, Risoluto, Multicom [45] and those from the Zhang lab), several efforts have been recently engaged towards developing end-to-end architectures. Here, we will shortly review these efforts and try to identify the key components of what represents end-to-end learning in protein structure prediction (Table 1).…”

“…The network is trained with a deep dilated resCNN to predict inter-residue distances directly from the raw MSA trRosetta [52] Computes traditional MSA features on the fly and passes them to dilated convolutional layers X-to-end learning NOVA [63] Adopts DeepFragLib from the same team which uses Long Short Term Memory units (LSTMs), to output a 3D structure DMPfold2 [49] The MSA, along with the precision matrix, is fed into a GRU, which outputs distances and angles (version used in CASP14) HMS-Casper [54] Raw sequences plus PSSMs are given to a "Recurrent Geometrical Network" comprising LSTM and geometric units and outputting a 3D structure a variety of of deep-learning models [91,16,92,93,21,38,73,42,41,37,45], including generative adversarial networks for contact map generation and refinement [94,35], got widely used to capture the same type of co-evolutionary information. One limitation of these methods is that they estimate average properties over an ensemble of sequences representative of a protein family.…”

Enhancing protein inter-residue real distance prediction by scrutinising deep learning models

“…Real valued distance prediction 16 , 21 has been addressed as a regression problem by Generative Adversarial Network-based method (GANProDist) 22 . Recent distance map prediction methods PDNET 23 and LiXu 24 (we name it after the author names since it has no original name) predict both real-valued and binned distances while another recent method DeepDist 25 predicts real-valued distances. Because of the vital role of distance maps in template-free or Free Modelling (FM) structure prediction, the Critical Assessment of protein Structure Prediction (CASP) organisers have introduced a new challenge category “inter-residue distance prediction” in CASP-14 26 .…”

“…State-of-the-art distance or contact map prediction algorithms 11 , 12 , 20 , 23 , 25 , 27 – 29 are largely based on Convolutional Neural Networks (CNN) 30 or Residual Networks (ResNet) 24 , 31 . Moreover, these methods predominantly use multiple sequence alignment (MSA) based coevolutionary features.…”

“…To overcome this, very recent distance prediction algorithms have used more and more coevolutionary and non-coevolutionary features and more complex neural networks. For example, PDNET 23 , DeepDist 25 , and LiXu 24 use respectively 3, 5, and 3 types of coevolutionary and 4, 7, and 3 types of non-coevolutionary features. Also, DeepDist 25 and LiXu 24 use ensembles of 4 and 6 ResNets respectively.…”