Class Prediction and Feature Selection with Linear Optimization for Metagenomic Count Data

The amount of metagenomic data is growing rapidly while the computational methods for metagenome analysis are still in their infancy. It is important to develop novel statistical learning tools for the prediction of associations between bacterial communities and disease phenotypes and for the detection of differentially abundant features. In this study, we presented a novel statistical learning method for simultaneous association prediction and feature selection with metagenomic samples from two or multiple treatment populations on the basis of count data. We developed a linear programming based support vector machine with and joint penalties for binary and multiclass classifications with metagenomic count data (metalinprog). We evaluated the performance of our method on several real and simulation datasets. The proposed method can simultaneously identify features and predict classes with the metagenomic count data.     

Most rare missense alleles are deleterious in humans: implications for complex disease and association studies

The accumulation of mildly deleterious missense mutations in individual human genomes has been proposed to be a genetic basis for complex diseases. The plausibility of this hypothesis depends on quantitative estimates of the prevalence of mildly deleterious de novo mutations and polymorphic variants in humans and on the intensity of selective pressure against them. We combined analysis of mutations causing human Mendelian diseases, of human-chimpanzee divergence, and of systematic data on human genetic variation and found that ~20% of new missense mutations in humans result in a loss of function, whereas ~27% are effectively neutral. Thus, the remaining 53% of new missense mutations have mildly deleterious effects. These mutations give rise to many low-frequency deleterious allelic variants in the human population, as is evident from a new data set of 37 genes sequenced in >1,500 individual human chromosomes. Surprisingly, up to 70% of low-frequency missense alleles are mildly deleterious and are associated with a heterozygous fitness loss in the range 0.001-0.003. Thus, the low allele frequency of an amino acid variant can, by itself, serve as a predictor of its functional significance. Several recent studies have reported a significant excess of rare missense variants in candidate genes or pathways in individuals with extreme values of quantitative phenotypes. These studies would be unlikely to yield results if most rare variants were neutral or if rare variants were not a significant contributor to the genetic component of phenotypic inheritance. Our results provide a justification for these types of candidate-gene (pathway) association studies and imply that mutation-selection balance may be a feasible evolutionary mechanism underlying some common diseases.     

Prediction of phenotypes of missense mutations in human proteins from biological assemblies

Single nucleotide polymorphisms (SNPs) are the most frequent variation in the human genome. Nonsynonymous SNPs that lead to missense mutations can be neutral or deleterious, and several computational methods have been presented that predict the phenotype of human missense mutations. These methods use sequence‐based and structure‐based features in various combinations, relying on different statistical distributions of these features for deleterious and neutral mutations. One structure‐based feature that has not been studied significantly is the accessible surface area within biologically relevant oligomeric assemblies. These assemblies are different from the crystallographic asymmetric unit for more than half of X‐ray crystal structures. We find that mutations in the core of proteins or in the interfaces in biological assemblies are significantly more likely to be disease‐associated than those on the surface of the biological assemblies. For structures with more than one protein in the biological assembly (whether the same sequence or different), we find the accessible surface area from biological assemblies provides a statistically significant improvement in prediction over the accessible surface area of monomers from protein crystal structures (P = 6e‐5). When adding this information to sequence‐based features such as the difference between wildtype and mutant position‐specific profile scores, the improvement from biological assemblies is statistically significant but much smaller (P = 0.018). Combining this information with sequence‐based features in a support vector machine leads to 82% accuracy on a balanced dataset of 50% disease‐associated mutations from SwissVar and 50% neutral mutations from human/primate sequence differences in orthologous proteins. Proteins 2013. © 2012 Wiley Periodicals, Inc.     

Predicting Genetic Variation Severity Using Machine Learning to Interpret Molecular Simulations

Distinct missense mutations in a specific gene have been associated with different diseases as well as differing severity of a disease. Current computational methods predict the potential pathogenicity of a missense variant but fail to differentiate between separate disease or severity phenotypes. We have developed a method to overcome this limitation by applying machine learning to features extracted from molecular dynamics simulations, creating a way to predict the effect of novel genetic variants in causing a disease, drug resistance, or another specific trait. As an example, we have applied this novel approach to variants in calmodulin associated with two distinct arrhythmias as well as two different neurodegenerative diseases caused by variants in amyloid-β peptide. The new method successfully predicts the specific disease caused by a gene variant and ranks its severity with more accuracy than existing methods. We call this method molecular dynamics phenotype prediction model.     

Feature Selection and the Class Imbalance Problem in Predicting Protein Function from Sequence

When the standard approach to predict protein function by sequence homology fails, other alternative methods can be used that require only the amino acid sequence for predicting function. One such approach uses machine learning to predict protein function directly from amino acid sequence features. However, there are two issues to consider before successful functional prediction can take place: identifying discriminatory features, and overcoming the challenge of a large imbalance in the training data. We show that by applying feature subset selection followed by undersampling of the majority class, significantly better support vector machine (SVM) classifiers are generated compared with standard machine learning approaches. As well as revealing that the features selected could have the potential to advance our understanding of the relationship between sequence and function, we also show that undersampling to produce fully balanced data significantly improves performance. The best discriminating ability is achieved using SVMs together with feature selection and full undersampling; this approach strongly outperforms other competitive learning algorithms. We conclude that this combined approach can generate powerful machine learning classifiers for predicting protein function directly from sequence.     

Adaptive weighted learning for unbalanced multicategory classification

Summary .  In multicategory classification, standard techniques typically treat all classes equally. This treatment can be problematic when the dataset is unbalanced in the sense that certain classes have very small class proportions compared to others. The minority classes may be ignored or discounted during the classification process due to their small proportions. This can be a serious problem if those minority classes are important. In this article, we study the problem of unbalanced classification and propose new criteria to measure classification accuracy. Moreover, we propose three different weighted learning procedures, two one-step weighted procedures, as well as one adaptive weighted procedure. We demonstrate the advantages of the new procedures, using multicategory support vector machines, through simulated and real datasets. Our results indicate that the proposed methodology can handle unbalanced classification problems effectively.     

Predicting deleterious nsSNPs: an analysis of sequence and structural attributes

Background  

There has been an explosion in the number of single nucleotide polymorphisms (SNPs) within public databases. In this study we focused on non-synonymous protein coding single nucleotide polymorphisms (nsSNPs), some associated with disease and others which are thought to be neutral. We describe the distribution of both types of nsSNPs using structural and sequence based features and assess the relative value of these attributes as predictors of function using machine learning methods. We also address the common problem of balance within machine learning methods and show the effect of imbalance on nsSNP function prediction. We show that nsSNP function prediction can be significantly improved by 100% undersampling of the majority class. The learnt rules were then applied to make predictions of function on all nsSNPs within Ensembl.     

基于基因组数据分析的细菌耐药基因识别与表型预测

利用基因组数据和生物信息学分析方法,快速鉴定耐药基因并预测耐药表型,为细菌耐药状况监测提供了有力辅助手段。目前,已有的数十个耐药数据库及其相关分析工具这些资源为细菌耐药基因的识别以及耐药表型的预测提供了数据信息和技术手段。随着细菌基因组数据的持续增加以及耐药表型数据的不断积累,大数据和机器学习能够更好地建立耐药表型与基因组信息之间的相关性,因此,构建高效的耐药表型预测模型成为研究热点。本文围绕细菌耐药基因的识别和耐药表型的预测,针对耐药相关数据库、耐药特征识别理论与方法、耐药数据的机器学习与表型预测等方面展开讨论,以期为细菌耐药的相关研究提供手段和思路。     

A joint fairness model with applications to risk predictions for underrepresented populations

In data collection for predictive modeling, underrepresentation of certain groups, based on gender, race/ethnicity, or age, may yield less accurate predictions for these groups. Recently, this issue of fairness in predictions has attracted significant attention, as data-driven models are increasingly utilized to perform crucial decision-making tasks. Existing methods to achieve fairness in the machine learning literature typically build a single prediction model in a manner that encourages fair prediction performance for all groups. These approaches have two major limitations: (i) fairness is often achieved by compromising accuracy for some groups; (ii) the underlying relationship between dependent and independent variables may not be the same across groups. We propose a joint fairness model (JFM) approach for logistic regression models for binary outcomes that estimates group-specific classifiers using a joint modeling objective function that incorporates fairness criteria for prediction. We introduce an accelerated smoothing proximal gradient algorithm to solve the convex objective function, and present the key asymptotic properties of the JFM estimates. Through simulations, we demonstrate the efficacy of the JFM in achieving good prediction performance and across-group parity, in comparison with the single fairness model, group-separate model, and group-ignorant model, especially when the minority group's sample size is small. Finally, we demonstrate the utility of the JFM method in a real-world example to obtain fair risk predictions for underrepresented older patients diagnosed with coronavirus disease 2019 (COVID-19).     

Testing computational prediction of missense mutation phenotypes: Functional characterization of 204 mutations of human cystathionine beta synthase

Predicting the phenotypes of missense mutations uncovered by large‐scale sequencing projects is an important goal in computational biology. High‐confidence predictions can be an aid in focusing experimental and association studies on those mutations most likely to be associated with causative relationships between mutation and disease. As an aid in developing these methods further, we have derived a set of random mutations of the enzymatic domains of human cystathionine beta synthase. This enzyme is a dimeric protein that catalyzes the condensation of serine and homocysteine to produce cystathionine. Yeast missing this enzyme cannot grow on medium lacking a source of cysteine, while transfection of functional human CBS into yeast strains missing endogenous enzyme can successfully complement for the missing gene. We used PCR mutagenesis with error‐prone Taq polymerase to produce 948 colonies and compared cell growth in the presence or absence of a cysteine source as a measure of CBS function. We were able to infer the phenotypes of 204 single‐site mutants, 79 of them deleterious and 125 neutral. This set was used to test the accuracy of six publicly available prediction methods for phenotype prediction of missense mutations: SIFT, PolyPhen, PMut, SNPs3D, PhD‐SNP, and nsSNPAnalyzer. The top methods are PolyPhen, SIFT, and nsSNPAnalyzer, which have similar performance. Using kernel discriminant functions, we found that the difference in position‐specific scoring matrix values is more predictive than the wild‐type PSSM score alone, and that the relative surface area in the biologically relevant complex is more predictive than that of the monomeric proteins. Proteins 2010. © 2010 Wiley‐Liss, Inc.     

A Bayesian ensemble approach with a disease gene network predicts damaging effects of missense variants of human cancers

Large-scale sequencing of cancer genomes has revealed many novel mutations and inter-tumoral heterogeneity. Therefore, prioritizing variants according to their potential deleterious effects has become essential. We constructed a disease gene network and proposed a Bayesian ensemble approach that integrates diverse sources to predict the functional effects of missense variants. We analyzed 23,336 missense disease mutations and 36,232 neutral polymorphisms of 12,039 human proteins. The results showed successful improvement of prediction accuracy in both sensitivity and specificity, and we demonstrated the utility of the method by applying it to somatic mutations obtained from colorectal and breast cancer cell lines. The candidate genes with predicted deleterious mutations as well as known cancer genes were significantly enriched in many KEGG pathways related to carcinogenesis, supporting genetic homogeneity of cancer at the pathway level. The breast cancer-specific network increased the prediction accuracy for breast cancer mutations. This study provides a ranked list of deleterious mutations and candidate cancer genes and suggests that mutations affecting cancer may occur in important pathways and should be interpreted on the phenotype-related network or pathway. A disease gene network may be of value in predicting functional effects of novel disease-specific mutations.     

The Impact of Neutral Mutations on Genome Evolvability

Beneficial mutations are rare and deleterious mutations are purged by natural selection. As a result, the vast majority of mutations that accumulate in genomes belong to the class of neutral mutations. Over the last two decades, neutral mutations, despite their null effect on fitness, have been shown to affect evolvability by providing access to new phenotypes through subsequent mutations that would not have been available otherwise. Here we propose that in addition, many mutations — independent of their selective effects — can affect the mutability of neighboring DNA sequences and modulate the efficacy of homologous recombination. Such mutations do not change the spectrum of accessible phenotypes, but rather the rate at which new phenotypes will be produced. Therefore, neutral mutations that accumulate in genomes have an important long-term impact on the evolutionary fate of genomes.     

A new disease-specific machine learning approach for the prediction of cancer-causing missense variants

High-throughput genotyping and sequencing techniques are rapidly and inexpensively providing large amounts of human genetic variation data. Single Nucleotide Polymorphisms (SNPs) are an important source of human genome variability and have been implicated in several human diseases, including cancer. Amino acid mutations resulting from non-synonymous SNPs in coding regions may generate protein functional changes that affect cell proliferation. In this study, we developed a machine learning approach to predict cancer-causing missense variants. We present a Support Vector Machine (SVM) classifier trained on a set of 3163 cancer-causing variants and an equal number of neutral polymorphisms. The method achieve 93% overall accuracy, a correlation coefficient of 0.86, and area under ROC curve of 0.98. When compared with other previously developed algorithms such as SIFT and CHASM our method results in higher prediction accuracy and correlation coefficient in identifying cancer-causing variants.     

Predicting Positive and Negative Relationships in Large Social Networks

In a social network, users hold and express positive and negative attitudes (e.g. support/opposition) towards other users. Those attitudes exhibit some kind of binary relationships among the users, which play an important role in social network analysis. However, some of those binary relationships are likely to be latent as the scale of social network increases. The essence of predicting latent binary relationships have recently began to draw researchers'' attention. In this paper, we propose a machine learning algorithm for predicting positive and negative relationships in social networks inspired by structural balance theory and social status theory. More specifically, we show that when two users in the network have fewer common neighbors, the prediction accuracy of the relationship between them deteriorates. Accordingly, in the training phase, we propose a segment-based training framework to divide the training data into two subsets according to the number of common neighbors between users, and build a prediction model for each subset based on support vector machine (SVM). Moreover, to deal with large-scale social network data, we employ a sampling strategy that selects small amount of training data while maintaining high accuracy of prediction. We compare our algorithm with traditional algorithms and adaptive boosting of them. Experimental results of typical data sets show that our algorithm can deal with large social networks and consistently outperforms other methods.     

Ensemble Positive Unlabeled Learning for Disease Gene Identification

An increasing number of genes have been experimentally confirmed in recent years as causative genes to various human diseases. The newly available knowledge can be exploited by machine learning methods to discover additional unknown genes that are likely to be associated with diseases. In particular, positive unlabeled learning (PU learning) methods, which require only a positive training set P (confirmed disease genes) and an unlabeled set U (the unknown candidate genes) instead of a negative training set N, have been shown to be effective in uncovering new disease genes in the current scenario. Using only a single source of data for prediction can be susceptible to bias due to incompleteness and noise in the genomic data and a single machine learning predictor prone to bias caused by inherent limitations of individual methods. In this paper, we propose an effective PU learning framework that integrates multiple biological data sources and an ensemble of powerful machine learning classifiers for disease gene identification. Our proposed method integrates data from multiple biological sources for training PU learning classifiers. A novel ensemble-based PU learning method EPU is then used to integrate multiple PU learning classifiers to achieve accurate and robust disease gene predictions. Our evaluation experiments across six disease groups showed that EPU achieved significantly better results compared with various state-of-the-art prediction methods as well as ensemble learning classifiers. Through integrating multiple biological data sources for training and the outputs of an ensemble of PU learning classifiers for prediction, we are able to minimize the potential bias and errors in individual data sources and machine learning algorithms to achieve more accurate and robust disease gene predictions. In the future, our EPU method provides an effective framework to integrate the additional biological and computational resources for better disease gene predictions.     

KFC2: a knowledge-based hot spot prediction method based on interface solvation, atomic density, and plasticity features

Hot spots constitute a small fraction of protein-protein interface residues, yet they account for a large fraction of the binding affinity. Based on our previous method (KFC), we present two new methods (KFC2a and KFC2b) that outperform other methods at hot spot prediction. A number of improvements were made in developing these new methods. First, we created a training data set that contained a similar number of hot spot and non-hot spot residues. In addition, we generated 47 different features, and different numbers of features were used to train the models to avoid over-fitting. Finally, two feature combinations were selected: One (used in KFC2a) is composed of eight features that are mainly related to solvent accessible surface area and local plasticity; the other (KFC2b) is composed of seven features, only two of which are identical to those used in KFC2a. The two models were built using support vector machines (SVM). The two KFC2 models were then tested on a mixed independent test set, and compared with other methods such as Robetta, FOLDEF, HotPoint, MINERVA, and KFC. KFC2a showed the highest predictive accuracy for hot spot residues (True Positive Rate: TPR = 0.85); however, the false positive rate was somewhat higher than for other models. KFC2b showed the best predictive accuracy for hot spot residues (True Positive Rate: TPR = 0.62) among all methods other than KFC2a, and the False Positive Rate (FPR = 0.15) was comparable with other highly predictive methods.     

Improving the assessment of the outcome of nonsynonymous SNVs with a consensus deleteriousness score, Condel

Several large ongoing initiatives that profit from next-generation sequencing technologies have driven--and in coming years will continue to drive--the emergence of long catalogs of missense single-nucleotide variants (SNVs) in the human genome. As a consequence, researchers have developed various methods and their related computational tools to classify these missense SNVs as probably deleterious or probably neutral polymorphisms. The outputs produced by each of these computational tools are of different natures and thus difficult to compare and integrate. Taking advantage of the possible complementarity between different tools might allow more accurate classifications. Here we propose an effective approach to integrating the output of some of these tools into a unified classification; this approach is based on a weighted average of the normalized scores of the individual methods (WAS). (In this paper, the approach is illustrated for the integration of five tools.) We show that this WAS outperforms each individual method in the task of classifying missense SNVs as deleterious or neutral. Furthermore, we demonstrate that this WAS can be used not only for classification purposes (deleterious versus neutral mutation) but also as an indicator of the impact of the mutation on the functionality of the mutant protein. In other words, it may be used as a deleteriousness score of missense SNVs. Therefore, we recommend the use of this WAS as a consensus deleteriousness score of missense mutations (Condel).     

Tumor classification using phylogenetic methods on expression data

Tumor classification is a well-studied problem in the field of bioinformatics. Developments in the field of DNA chip design have now made it possible to measure the expression levels of thousands of genes in sample tissue from healthy cell lines or tumors. A number of studies have examined the problems of tumor classification: class discovery, the problem of defining a number of classes of tumors using the data from a DNA chip, and class prediction, the problem of accurately classifying an unknown tumor, given expression data from the unknown tumor and from a learning set. The current work has applied phylogenetic methods to both problems. To solve the class discovery problem, we impose a metric on a set of tumors as a function of their gene expression levels, and impose a tree structure on this metric, using standard tree fitting methods borrowed from the field of phylogenetics. Phylogenetic methods provide a simple way of imposing a clear hierarchical relationship on the data, with branch lengths in the classification tree representing the degree of separation witnessed. We tested our method for class discovery on two data sets: a data set of 87 tissues, comprised mostly of small, round, blue-cell tumors (SRBCTs), and a data set of 22 breast tumors. We fit the 87 samples of the first set to a classification tree, which neatly separated into four major clusters corresponding exactly to the four groups of tumors, namely neuroblastomas, rhabdomyosarcomas, Burkitt's lymphomas, and the Ewing's family of tumors. The classification tree built using the breast cancer data separated tumors with BRCA1 mutations from those with BRCA2 mutations, with sporadic tumors separated from both groups and from each other. We also demonstrate the flexibility of the class discovery method with regard to standard resampling methodology such as jackknifing and noise perturbation. To solve the class prediction problem, we built a classification tree on the learning set, and then sought the optimal placement of each test sample within the classification tree. We tested this method on the SRBCT data set, and classified each tumor successfully.     

Predicting gene function using few positive examples and unlabeled ones

Background

A large amount of functional genomic data have provided enough knowledge in predicting gene function computationally, which uses known functional annotations and relationship between unknown genes and known ones to map unknown genes to GO functional terms. The prediction procedure is usually formulated as binary classification problem. Training binary classifier needs both positive examples and negative ones that have almost the same size. However, from various annotation database, we can only obtain few positive genes annotation for most offunctional terms, that is, there are only few positive examples for training classifier, which makes predicting directly gene function infeasible.

Results

We propose a novel approach SPE_RNE to train classifier for each functional term. Firstly, positive examples set is enlarged by creating synthetic positive examples. Secondly, representative negative examples are selected by training SVM(support vector machine) iteratively to move classification hyperplane to a appropriate place. Lastly, an optimal SVM classifier are trained by using grid search technique. On combined kernel ofYeast protein sequence, microarray expression, protein-protein interaction and GO functional annotation data, we compare SPE_RNE with other three typical methods in three classical performance measures recall R, precise P and their combination F: twoclass considers all unlabeled genes as negative examples, twoclassbal selects randomly same number negative examples from unlabeled gene, PSoL selects a negative examples set that are far from positive examples and far from each other.

Conclusions

In test data and unknown genes data, we compute average and variant of measure F. The experiments showthat our approach has better generalized performance and practical prediction capacity. In addition, our method can also be used for other organisms such as human.

     

环境微生物研究中机器学习算法及应用

微生物在环境中无处不在，它们不仅是生物地球化学循环和环境演化的关键参与者，也在环境监测、生态治理和保护中发挥着重要作用。随着高通量技术的发展，大量微生物数据产生，运用机器学习对环境微生物大数据进行建模和分析，在微生物标志物识别、污染物预测和环境质量预测等领域的科学研究和社会应用方面均具有重要意义。机器学习可分为监督学习和无监督学习2大类。在微生物组学研究当中，无监督学习通过聚类、降维等方法高效地学习输入数据的特征，进而对微生物数据进行整合和归类。监督学习运用有特征和标记的微生物数据集训练模型，在面对只有特征没有标记的数据时可以判断出标记，从而实现对新数据的分类、识别和预测。然而，复杂的机器学习算法通常以牺牲可解释性为代价来重点关注模型预测的准确性。机器学习模型通常可以看作预测特定结果的“黑匣子”，即对模型如何得出预测所知甚少。为了将机器学习更多地运用于微生物组学研究、提高我们提取有价值的微生物信息的能力，深入了解机器学习算法、提高模型的可解释性尤为重要。本文主要介绍在环境微生物领域常用的机器学习算法和基于微生物组数据的机器学习模型的构建步骤，包括特征选择、算法选择、模型构建和评估等，并对各种机器学习模型在环境微生物领域的应用进行综述，深入探究微生物组与周围环境之间的关联，探讨提高模型可解释性的方法，并为未来环境监测、环境健康预测提供科学参考。