Inspired by empirical studies of networked systems such as the Internet, social networks, and biological networks, researchers have in recent years developed a variety of techniques and models to help us understand or predict the behavior of these systems. Here we review developments in this field, including such concepts as the small-world effect, degree distributions, clustering, network correlations, random graph models, models of network growth and preferential attachment, and dynamical processes taking place on networks.

/pdf/the-structure-and-function-of-complex-networks-34deompc9l.pdf

The Structure and Function of Complex Networks

Statistical method for testing the neutral mutation hypothesis by DNA polymorphism.

Proteins are essential to life, and understanding their structure can facilitate a mechanistic understanding of their function. Through an enormous experimental effort1–4, the structures of around 100,000 unique proteins have been determined5, but this represents a small fraction of the billions of known protein sequences6,7. Structural coverage is bottlenecked by the months to years of painstaking effort required to determine a single protein structure. Accurate computational approaches are needed to address this gap and to enable large-scale structural bioinformatics. Predicting the three-dimensional structure that a protein will adopt based solely on its amino acid sequence—the structure prediction component of the ‘protein folding problem’8—has been an important open research problem for more than 50 years9. Despite recent progress10–14, existing methods fall far short of atomic accuracy, especially when no homologous structure is available. Here we provide the first computational method that can regularly predict protein structures with atomic accuracy even in cases in which no similar structure is known. We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Critical Assessment of protein Structure Prediction (CASP14)15, demonstrating accuracy competitive with experimental structures in a majority of cases and greatly outperforming other methods. Underpinning the latest version of AlphaFold is a novel machine learning approach that incorporates physical and biological knowledge about protein structure, leveraging multi-sequence alignments, into the design of the deep learning algorithm. AlphaFold predicts protein structures with an accuracy competitive with experimental structures in the majority of cases using a novel deep learning architecture.

/pdf/highly-accurate-protein-structure-prediction-with-alphafold-2xjbzfwu3n.pdf

Highly accurate protein structure prediction with AlphaFold

Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

http://www.maths.qmul.ac.uk/%7Elatora/report_06.pdf

Complex networks: Structure and dynamics

Profiling developmental toxicity of 387 environmental chemicals using EPA's Toxicity Reference Database (ToxRefDB)

A method is described for folding polymers to specific target conformations. The approach uses a fast but approximate dynamics algorithm, coupled with a genetic algorithm that is used to evolve the large number of free parameters needed. The dynamics algorithm uses a state transition matrix approach. At each time step, the distances between pairs of atoms are adjusted by shifting them from Dij to Dij + Sij where Sij is an element of the state transition matrix S. Atom pairs that are attractive have Sij 0. The atomic movement is carried out by gradient minimizing the molecular mechanics energy of the molecule subject to harmonic distance constraints. The method is applied to a simple test case, a 19 atom 2-D polymer. The paper also show that the S matrices can correctly fold a limited variety of initial conformations that differ from the one used during the evolution phase.

Computational evolution of a model polymer that folds to a specified target conformation.

REACH substances DTU Orbit (07/12/2018) Development of a QSAR Model for Thyroperoxidase Inhibition and Screening of 72,526 REACH substances Thyroid hormones (THs) are involved in multiple biological processes and are critical modulators of fetal development. Even moderate changes in maternal or fetal TH levels can produce irreversible neurological deficits in children, such as lower IQ. The enzyme thyroperoxidase (TPO) plays a key role in the synthesis of THs, and inhibition of TPO by xenobiotics results in decreased TH synthesis. Recently, a high-throughput screening assay for TPO inhibition (AUR-TPO) was developed and used to test the ToxCast Phase I and II chemicals. In the present study, we used the results from AUR-TPO to develop a Quantitative Structure Activity Relationship (QSAR) model for TPO inhibition. The training set consisted of 898 discrete organic chemicals: 134 inhibitors and 764 non-inhibitors. A five times two-fold cross-validation of the model was performed, yielding a balanced accuracy of 78.7%. More recently, an additional ~800 chemicals were tested in the AUR-TPO assay. These data were used for a blinded external validation of the QSAR model, demonstrating a balanced accuracy of 85.7%. Overall, the crossand external validation indicate a robust model with high predictive performance. Next, we used the QSAR model to predict 72,526 REACH pre registered substances. The model could predict 49.5% (35,925) of the substances in its applicability domain and of these, 8,863 (24.7%) were predicted to be TPO inhibitors. Predictions from this screening can be used in a tiered approach to prioritize potential thyroid disrupting chemical substances for further evaluation.

Development of a QSAR Model for Thyroperoxidase Inhibition and Screening of 72,526 REACH substances

Genome-wide association of haplotype markers to gene expression levels.

Poster presented at the American Society for Cellular and Computational Toxicology (ASCCT) annual meeting in September 2019

Richard S. Judson

Papers

Profiling developmental toxicity of 387 environmental chemicals using EPA's Toxicity Reference Database (ToxRefDB)

Computational evolution of a model polymer that folds to a specified target conformation.

Development of a QSAR Model for Thyroperoxidase Inhibition and Screening of 72,526 REACH substances

Genome-wide association of haplotype markers to gene expression levels.

Evaluation of the Predictive Accuracy of QSAR Models and Alerts for Genotoxicity Using a Newly Compiled Experimental Dataset