Maximum-likelihood crystallization

The crystallization facility of the TB Structural Genomics Consortium, one of nine NIH-sponsored structural genomics pilot projects, employs a combinatorial random sampling technique in high-throughput crystallization screening. Although data are still sparse and a comprehensive analysis cannot be performed at this stage, preliminary results appear to validate the random-screening concept. A discussion of statistical crystallization data analysis aims to draw attention to the need for comprehensive and valid sampling protocols. In view of limited overlap in techniques and sampling parameters between the publicly funded high-throughput crystallography initiatives, exchange of information should be encouraged, aiming to effectively integrate data mining efforts into a comprehensive predictive framework for protein crystallization.     

RIBER/DIBER: a software suite for crystal content analysis in the studies of protein-nucleic acid complexes

Co-crystallization experiments of proteins with nucleic acids do not guarantee that both components are present in the crystal. We have previously developed DIBER to predict crystal content when protein and DNA are present in the crystallization mix. Here, we present RIBER, which should be used when protein and RNA are in the crystallization drop. The combined RIBER/DIBER suite builds on machine learning techniques to make reliable, quantitative predictions of crystal content for non-expert users and high-throughput crystallography.     

PredPPCrys: Accurate Prediction of Sequence Cloning,Protein Production,Purification and Crystallization Propensity from Protein Sequences Using Multi-Step Heterogeneous Feature Fusion and Selection

X-ray crystallography is the primary approach to solve the three-dimensional structure of a protein. However, a major bottleneck of this method is the failure of multi-step experimental procedures to yield diffraction-quality crystals, including sequence cloning, protein material production, purification, crystallization and ultimately, structural determination. Accordingly, prediction of the propensity of a protein to successfully undergo these experimental procedures based on the protein sequence may help narrow down laborious experimental efforts and facilitate target selection. A number of bioinformatics methods based on protein sequence information have been developed for this purpose. However, our knowledge on the important determinants of propensity for a protein sequence to produce high diffraction-quality crystals remains largely incomplete. In practice, most of the existing methods display poorer performance when evaluated on larger and updated datasets. To address this problem, we constructed an up-to-date dataset as the benchmark, and subsequently developed a new approach termed ‘PredPPCrys’ using the support vector machine (SVM). Using a comprehensive set of multifaceted sequence-derived features in combination with a novel multi-step feature selection strategy, we identified and characterized the relative importance and contribution of each feature type to the prediction performance of five individual experimental steps required for successful crystallization. The resulting optimal candidate features were used as inputs to build the first-level SVM predictor (PredPPCrys I). Next, prediction outputs of PredPPCrys I were used as the input to build second-level SVM classifiers (PredPPCrys II), which led to significantly enhanced prediction performance. Benchmarking experiments indicated that our PredPPCrys method outperforms most existing procedures on both up-to-date and previous datasets. In addition, the predicted crystallization targets of currently non-crystallizable proteins were provided as compendium data, which are anticipated to facilitate target selection and design for the worldwide structural genomics consortium. PredPPCrys is freely available at http://www.structbioinfor.org/PredPPCrys.     

Models of protein crystal growth

The growth of large and well ordered protein crystals remains the major obstacle in protein structure determination by means of X-ray crystallography. One of the reasons is that the physico-chemical aspect of protein crystallization process is not understood. This article reviews efforts towards formulation of models that could become theoretical frameworks for the interpretation of voluminous experimental data collected on protein crystal growth. Special attention is devoted to microscopic models that recognize the role of the shape of protein molecules in crystal formation.     

Machine Learning: How Much Does It Tell about Protein Folding Rates?

The prediction of protein folding rates is a necessary step towards understanding the principles of protein folding. Due to the increasing amount of experimental data, numerous protein folding models and predictors of protein folding rates have been developed in the last decade. The problem has also attracted the attention of scientists from computational fields, which led to the publication of several machine learning-based models to predict the rate of protein folding. Some of them claim to predict the logarithm of protein folding rate with an accuracy greater than 90%. However, there are reasons to believe that such claims are exaggerated due to large fluctuations and overfitting of the estimates. When we confronted three selected published models with new data, we found a much lower predictive power than reported in the original publications. Overly optimistic predictive powers appear from violations of the basic principles of machine-learning. We highlight common misconceptions in the studies claiming excessive predictive power and propose to use learning curves as a safeguard against those mistakes. As an example, we show that the current amount of experimental data is insufficient to build a linear predictor of logarithms of folding rates based on protein amino acid composition.     

Modeling Dynamic Systems with Efficient Ensembles of Process-Based Models

Ensembles are a well established machine learning paradigm, leading to accurate and robust models, predominantly applied to predictive modeling tasks. Ensemble models comprise a finite set of diverse predictive models whose combined output is expected to yield an improved predictive performance as compared to an individual model. In this paper, we propose a new method for learning ensembles of process-based models of dynamic systems. The process-based modeling paradigm employs domain-specific knowledge to automatically learn models of dynamic systems from time-series observational data. Previous work has shown that ensembles based on sampling observational data (i.e., bagging and boosting), significantly improve predictive performance of process-based models. However, this improvement comes at the cost of a substantial increase of the computational time needed for learning. To address this problem, the paper proposes a method that aims at efficiently learning ensembles of process-based models, while maintaining their accurate long-term predictive performance. This is achieved by constructing ensembles with sampling domain-specific knowledge instead of sampling data. We apply the proposed method to and evaluate its performance on a set of problems of automated predictive modeling in three lake ecosystems using a library of process-based knowledge for modeling population dynamics. The experimental results identify the optimal design decisions regarding the learning algorithm. The results also show that the proposed ensembles yield significantly more accurate predictions of population dynamics as compared to individual process-based models. Finally, while their predictive performance is comparable to the one of ensembles obtained with the state-of-the-art methods of bagging and boosting, they are substantially more efficient.     

Machine learning approaches for prediction of linear B-cell epitopes on proteins

Identification and characterization of antigenic determinants on proteins has received considerable attention utilizing both, experimental as well as computational methods. For computational routines mostly structural as well as physicochemical parameters have been utilized for predicting the antigenic propensity of protein sites. However, the performance of computational routines has been low when compared to experimental alternatives. Here we describe the construction of machine learning based classifiers to enhance the prediction quality for identifying linear B-cell epitopes on proteins. Our approach combines several parameters previously associated with antigenicity, and includes novel parameters based on frequencies of amino acids and amino acid neighborhood propensities. We utilized machine learning algorithms for deriving antigenicity classification functions assigning antigenic propensities to each amino acid of a given protein sequence. We compared the prediction quality of the novel classifiers with respect to established routines for epitope scoring, and tested prediction accuracy on experimental data available for HIV proteins. The major finding is that machine learning classifiers clearly outperform the reference classification systems on the HIV epitope validation set.     

CRYSpred: accurate sequence-based protein crystallization propensity prediction using sequence-derived structural characteristics

Relatively low success rates of X-ray crystallography, which is the most popular method for solving proteins structures, motivate development of novel methods that support selection of tractable protein targets. This aspect is particularly important in the context of the current structural genomics efforts that allow for a certain degree of flexibility in the target selection. We propose CRYSpred, a novel in-silico crystallization propensity predictor that uses a set of 15 novel features which utilize a broad range of inputs including charge, hydrophobicity, and amino acid composition derived from the protein chain, and the solvent accessibility and disorder predicted from the protein sequence. Our method outperforms seven modern crystallization propensity predictors on three, independent from training dataset, benchmark test datasets. The strong predictive performance offered by the CRYSpred is attributed to the careful design of the features, utilization of the comprehensive set of inputs, and the usage of the Support Vector Machine classifier. The inputs utilized by CRYSpred are well-aligned with the existing rules-of-thumb that are used in the structural genomics studies.     

Machine learning modeling of family wide enzyme-substrate specificity screens

Biocatalysis is a promising approach to sustainably synthesize pharmaceuticals, complex natural products, and commodity chemicals at scale. However, the adoption of biocatalysis is limited by our ability to select enzymes that will catalyze their natural chemical transformation on non-natural substrates. While machine learning and in silico directed evolution are well-posed for this predictive modeling challenge, efforts to date have primarily aimed to increase activity against a single known substrate, rather than to identify enzymes capable of acting on new substrates of interest. To address this need, we curate 6 different high-quality enzyme family screens from the literature that each measure multiple enzymes against multiple substrates. We compare machine learning-based compound-protein interaction (CPI) modeling approaches from the literature used for predicting drug-target interactions. Surprisingly, comparing these interaction-based models against collections of independent (single task) enzyme-only or substrate-only models reveals that current CPI approaches are incapable of learning interactions between compounds and proteins in the current family level data regime. We further validate this observation by demonstrating that our no-interaction baseline can outperform CPI-based models from the literature used to guide the discovery of kinase inhibitors. Given the high performance of non-interaction based models, we introduce a new structure-based strategy for pooling residue representations across a protein sequence. Altogether, this work motivates a principled path forward in order to build and evaluate meaningful predictive models for biocatalysis and other drug discovery applications.     

An interaction network predicted from public data as a discovery tool: application to the Hsp90 molecular chaperone machine

Understanding the functions of proteins requires information about their protein-protein interactions (PPI). The collective effort of the scientific community generates far more data on any given protein than individual experimental approaches. The latter are often too limited to reveal an interactome comprehensively. We developed a workflow for parallel mining of all major PPI databases, containing data from several model organisms, and to integrate data from the literature for a protein of interest. We applied this novel approach to build the PPI network of the human Hsp90 molecular chaperone machine (Hsp90Int) for which previous efforts have yielded limited and poorly overlapping sets of interactors. We demonstrate the power of the Hsp90Int database as a discovery tool by validating the prediction that the Hsp90 co-chaperone Aha1 is involved in nucleocytoplasmic transport. Thus, we both describe how to build a custom database and introduce a powerful new resource for the scientific community.     

DNA4mcEL：基于核苷酸信息特征计算分析与预测

N⁴-甲基胞嘧啶(N4-methylcytosine, 4mC)是一种重要的表观遗传修饰,在DNA的修复、表达和复制中发挥重要作用。准确鉴定4mC位点有助于深入研究其生物学功能和机制,由于4mC位点的实验鉴定即耗时又昂贵,特别是考虑到基因序列的快速积累,迫切需要补充有效的计算方法。因此,提供一个快速、准确的4mC位点在线预测平台十分必要。目前,还未见对构建必要的预测模型所需的不同特征的机器学习（machine learning, ML）方法进行全面的分析和评估。我们构建多组特征集,并且采用5种ML方法（如随机森林,支持向量机,集成学习等）,提出一种称为“DNA4mcEL”的预测方法。在随机10折交叉验证测试下与现有的预测器相比,DNA4mcEL预测C. elegans、D. melanogaster、A. thaliana、E. coli、G. subterraneus、G. pickeringii 6个物种的精度均有提高。基于本方法的预测器DNA4mcEL在这项任务中显著优于现有的预测器。我们希望通过这个综合调查和建立更准确模型的策略,可以作为激发N⁴-甲基胞嘧啶预测计算方法未来发展的有用指南,加快新N⁴-甲基胞嘧啶的发现。DNA4mcEL的独立版本可以从https://github.com/kukuky00/DNA4mcEL.git免费获得。     

Novel machine learning approaches revolutionize protein knowledge

Breakthrough methods in machine learning (ML), protein structure prediction, and novel ultrafast structural aligners are revolutionizing structural biology. Obtaining accurate models of proteins and annotating their functions on a large scale is no longer limited by time and resources. The most recent method to be top ranked by the Critical Assessment of Structure Prediction (CASP) assessment, AlphaFold 2 (AF2), is capable of building structural models with an accuracy comparable to that of experimental structures. Annotations of 3D models are keeping pace with the deposition of the structures due to advancements in protein language models (pLMs) and structural aligners that help validate these transferred annotations. In this review we describe how recent developments in ML for protein science are making large-scale structural bioinformatics available to the general scientific community.     

Prediction Errors in Learning Drug Response from Gene Expression Data – Influence of Labeling,Sample Size,and Machine Learning Algorithm

Model-based prediction is dependent on many choices ranging from the sample collection and prediction endpoint to the choice of algorithm and its parameters. Here we studied the effects of such choices, exemplified by predicting sensitivity (as IC50) of cancer cell lines towards a variety of compounds. For this, we used three independent sample collections and applied several machine learning algorithms for predicting a variety of endpoints for drug response. We compared all possible models for combinations of sample collections, algorithm, drug, and labeling to an identically generated null model. The predictability of treatment effects varies among compounds, i.e. response could be predicted for some but not for all. The choice of sample collection plays a major role towards lowering the prediction error, as does sample size. However, we found that no algorithm was able to consistently outperform the other and there was no significant difference between regression and two- or three class predictors in this experimental setting. These results indicate that response-modeling projects should direct efforts mainly towards sample collection and data quality, rather than method adjustment.     

Mutation probability of cytochrome P450 based on a genetic algorithm and support vector machine

The support vector machine (SVM), an effective statistical learning method, has been widely used in mutation prediction. Two factors, i.e., feature selection and parameter setting, have shown great influence on the efficiency and accuracy of SVM classification. In this study, according to the principles of a genetic algorithm (GA) and SVM, we developed a GA-SVM program and applied it to human cytochrome P450s (CYP450s), which are important monooxygenases in phase I drug metabolism. The program optimizes features and parameters simultaneously, and hence fewer features are used and the overall prediction accuracy is improved. We focus on the mutation of non-synonymous single nucleotide polymorphisms (nsSNPs) in protein sequences that appear to exhibit significant influences on drug metabolism. The final predictive model has a quite satisfactory performance, with the prediction accuracy of 61% and cross-validation accuracy of 73%. The results indicate that the GA-SVM program is a powerful tool in optimizing mutation predictive models of nsSNPs of human CYP450s.     

An Approach Based on Mutually Informed Neural Networks to Optimize the Generalization Capabilities of Decision Support Systems Developed for Heart Failure Prediction

Available clinical methods for heart failure (HF) diagnosis are expensive and require a high-level of experts intervention. Recently, various machine learning models have been developed for the prediction of HF where most of them have an issue of over-fitting. Over-fitting occurs when machine learning based predictive models show better performance on the training data yet demonstrate a poor performance on the testing data and the other way around. Developing a machine learning model which is able to produce generalization capabilities (such that the model exhibits better performance on both the training and the testing data sets) could overall minimize the prediction errors. Hence, such prediction models could potentially be helpful to cardiologists for the effective diagnose of HF. This paper proposes a two-stage decision support system to overcome the over-fitting issue and to optimize the generalization factor. The first stage uses a mutual information based statistical model while the second stage uses a neural network. We applied our approach to the HF subset of publicly available Cleveland heart disease database. Our experimental results show that the proposed decision support system has optimized the generalization capabilities and has reduced the mean percent error (MPE) to 8.8% which is significantly less than the recently published studies. In addition, our model exhibits a 93.33% accuracy rate which is higher than twenty eight recently developed HF risk prediction models that achieved accuracy in the range of 57.85% to 92.31%. We can hope that our decision support system will be helpful to cardiologists if deployed in clinical setup.     

The promise of machine learning in predicting treatment outcomes in psychiatry

For many years, psychiatrists have tried to understand factors involved in response to medications or psychotherapies, in order to personalize their treatment choices. There is now a broad and growing interest in the idea that we can develop models to personalize treatment decisions using new statistical approaches from the field of machine learning and applying them to larger volumes of data. In this pursuit, there has been a paradigm shift away from experimental studies to confirm or refute specific hypotheses towards a focus on the overall explanatory power of a predictive model when tested on new, unseen datasets. In this paper, we review key studies using machine learning to predict treatment outcomes in psychiatry, ranging from medications and psychotherapies to digital interventions and neurobiological treatments. Next, we focus on some new sources of data that are being used for the development of predictive models based on machine learning, such as electronic health records, smartphone and social media data, and on the potential utility of data from genetics, electrophysiology, neuroimaging and cognitive testing. Finally, we discuss how far the field has come towards implementing prediction tools in real‐world clinical practice. Relatively few retrospective studies to‐date include appropriate external validation procedures, and there are even fewer prospective studies testing the clinical feasibility and effectiveness of predictive models. Applications of machine learning in psychiatry face some of the same ethical challenges posed by these techniques in other areas of medicine or computer science, which we discuss here. In short, machine learning is a nascent but important approach to improve the effectiveness of mental health care, and several prospective clinical studies suggest that it may be working already.     

Systems Metabolic Engineering Meets Machine Learning: A New Era for Data‐Driven Metabolic Engineering

The recent increase in high‐throughput capacity of ‘omics datasets combined with advances and interest in machine learning (ML) have created great opportunities for systems metabolic engineering. In this regard, data‐driven modeling methods have become increasingly valuable to metabolic strain design. In this review, the nature of ‘omics is discussed and a broad introduction to the ML algorithms combining these datasets into predictive models of metabolism and metabolic rewiring is provided. Next, this review highlights recent work in the literature that utilizes such data‐driven methods to inform various metabolic engineering efforts for different classes of application including product maximization, understanding and profiling phenotypes, de novo metabolic pathway design, and creation of robust system‐scale models for biotechnology. Overall, this review aims to highlight the potential and promise of using ML algorithms with metabolic engineering and systems biology related datasets.     

CrowdGO: Machine learning and semantic similarity guided consensus Gene Ontology annotation

Characterising gene function for the ever-increasing number and diversity of species with annotated genomes relies almost entirely on computational prediction methods. These software are also numerous and diverse, each with different strengths and weaknesses as revealed through community benchmarking efforts. Meta-predictors that assess consensus and conflict from individual algorithms should deliver enhanced functional annotations. To exploit the benefits of meta-approaches, we developed CrowdGO, an open-source consensus-based Gene Ontology (GO) term meta-predictor that employs machine learning models with GO term semantic similarities and information contents. By re-evaluating each gene-term annotation, a consensus dataset is produced with high-scoring confident annotations and low-scoring rejected annotations. Applying CrowdGO to results from a deep learning-based, a sequence similarity-based, and two protein domain-based methods, delivers consensus annotations with improved precision and recall. Furthermore, using standard evaluation measures CrowdGO performance matches that of the community’s best performing individual methods. CrowdGO therefore offers a model-informed approach to leverage strengths of individual predictors and produce comprehensive and accurate gene functional annotations.     

The potential of random forest and neural networks for biomass and recombinant protein modeling in Escherichia coli fed‐batch fermentations

Product quality assurance strategies in production of biopharmaceuticals currently undergo a transformation from empirical “quality by testing” to rational, knowledge‐based “quality by design” approaches. The major challenges in this context are the fragmentary understanding of bioprocesses and the severely limited real‐time access to process variables related to product quality and quantity. Data driven modeling of process variables in combination with model predictive process control concepts represent a potential solution to these problems. The selection of statistical techniques best qualified for bioprocess data analysis and modeling is a key criterion. In this work a series of recombinant Escherichia coli fed‐batch production processes with varying cultivation conditions employing a comprehensive on‐ and offline process monitoring platform was conducted. The applicability of two machine learning methods, random forest and neural networks, for the prediction of cell dry mass and recombinant protein based on online available process parameters and two‐dimensional multi‐wavelength fluorescence spectroscopy is investigated. Models solely based on routinely measured process variables give a satisfying prediction accuracy of about ± 4% for the cell dry mass, while additional spectroscopic information allows for an estimation of the protein concentration within ± 12%. The results clearly argue for a combined approach: neural networks as modeling technique and random forest as variable selection tool.