Inferring causal phenotype networks from segregating populations

A major goal in the study of complex traits is to decipher the causal interrelationships among correlated phenotypes. Current methods mostly yield undirected networks that connect phenotypes without causal orientation. Some of these connections may be spurious due to partial correlation that is not causal. We show how to build causal direction into an undirected network of phenotypes by including causal QTL for each phenotype. We evaluate causal direction for each edge connecting two phenotypes, using a LOD score. This new approach can be applied to many different population structures, including inbred and outbred crosses as well as natural populations, and can accommodate feedback loops. We assess its performance in simulation studies and show that our method recovers network edges and infers causal direction correctly at a high rate. Finally, we illustrate our method with an example involving gene expression and metabolite traits from experimental crosses.     

Mitochondrial DNA haplogroup D4a is a marker for extreme longevity in Japan

We report results from the analysis of complete mitochondrial DNA (mtDNA) sequences from 112 Japanese semi-supercentenarians (aged above 105 years) combined with previously published data from 96 patients in each of three non-disease phenotypes: centenarians (99-105 years of age), healthy non-obese males, obese young males and four disease phenotypes, diabetics with and without angiopathy, and Alzheimer's and Parkinson's disease patients. We analyze the correlation between mitochondrial polymorphisms and the longevity phenotype using two different methods. We first use an exhaustive algorithm to identify all maximal patterns of polymorphisms shared by at least five individuals and define a significance score for enrichment of the patterns in each phenotype relative to healthy normals. Our study confirms the correlations observed in a previous study showing enrichment of a hierarchy of haplogroups in the D clade for longevity. For the extreme longevity phenotype we see a single statistically significant signal: a progressive enrichment of certain "beneficial" patterns in centenarians and semi-supercentenarians in the D4a haplogroup. We then use Principal Component Spectral Analysis of the SNP-SNP Covariance Matrix to compare the measured eigenvalues to a Null distribution of eigenvalues on Gaussian datasets to determine whether the correlations in the data (due to longevity) arises from some property of the mutations themselves or whether they are due to population structure. The conclusion is that the correlations are entirely due to population structure (phylogenetic tree). We find no signal for a functional mtDNA SNP correlated with longevity. The fact that the correlations are from the population structure suggests that hitch-hiking on autosomal events is a possible explanation for the observed correlations.     

Nonidentity of ribosomal structural proteins in growing and starved Tetrahymena

We have examined the ribosomal structural proteins isolated from vegetatively growing Tetrahymena pyriformis and from cells that had been starved of all nutrients for 24 h. Reproducible, nonartifactual differences in protein complement, primarily associated with the large ribosomal subunit, were found. The kinetics of change in ribosomal protein complement were followed both in refed and in newly starved cells. Furthermore, attempts at correlating a certain protein "phenotype" with a particular functional state of the ribosome were made. It was concluded that the alterations seen could not be correlated with a specific stage in the normal ribosome cycle. We did show, however, that the change in protein complement could occur as a result of altering preexisting ribosomes. In addition, we showed that the change correlates with a decrease in growth rate rather than being caused by the starvation conditions themselves. Speculations as to the functional significance of the protein changes are presented.     

Coronary Injury Score Correlates with Proliferating Cells and Alpha-Smooth Muscle Actin Expression in Stented Porcine Coronary Arteries

Neointimal formation and cell proliferation resulting into in-stent restenosis is a major pathophysiological event following the deployment of stents in the coronary arteries. In this study, we assessed the degree of injury, based on damage to internal elastic lamina, media, external elastic lamina, and adventitia following the intravascular stenting, and its relationship with the degree of smooth muscle cell proliferation. We examined the smooth muscle cell proliferation and their phenotype at different levels of stent injury in the coronary arteries of domestic swine fed a normal swine diet. Five weeks after stent implantation, swine with and without stents were euthanized and coronaries were excised. Arteries were embedded in methyl methacrylate and sections were stained with H&E, trichrome, and Movat’s pentachrome. The expression of Ki67, α-smooth muscle actin (SMA), vimentin, and HMGB1 was evaluated by immunofluorescence. There was a positive correlation between percent area stenosis and injury score. The distribution of SMA and vimentin was correlated with the degree of arterial injury such that arteries that had an injury score >2 did not have immunoreactivity to SMA in the neointimal cells near the stent struts, but these neointimal cells were positive for vimentin, suggesting a change in the smooth muscle cell phenotype. The Ki67 and HMGB1 immunoreactivity was highly correlated with the fragmentation of the IEL and injury in the tunica media. Thus, the extent of coronary arterial injury during interventional procedure will dictate the degree of neointimal hyperplasia, in-stent restenosis, and smooth muscle cell phenotype.     

Quantitative trait associated microarray gene expression data analysis

Selection on phenotypes may cause genetic change. To understand the relationship between phenotype and gene expression from an evolutionary viewpoint, it is important to study the concordance between gene expression and profiles of phenotypes. In this study, we use a novel method of clustering to identify genes whose expression profiles are related to a quantitative phenotype. Cluster analysis of gene expression data aims at classifying genes into several different groups based on the similarity of their expression profiles across multiple conditions. The hope is that genes that are classified into the same clusters may share underlying regulatory elements or may be a part of the same metabolic pathways. Current methods for examining the association between phenotype and gene expression are limited to linear association measured by the correlation between individual gene expression values and phenotype. Genes may be associated with the phenotype in a nonlinear fashion. In addition, groups of genes that share a particular pattern in their relationship to phenotype may be of evolutionary interest. In this study, we develop a method to group genes based on orthogonal polynomials under a multivariate Gaussian mixture model. The effect of each expressed gene on the phenotype is partitioned into a cluster mean and a random deviation from the mean. Genes can also be clustered based on a time series. Parameters are estimated using the expectation-maximization algorithm and implemented in SAS. The method is verified with simulated data and demonstrated with experimental data from 2 studies, one clusters with respect to severity of disease in Alzheimer's patients and another clusters data for a rat fracture healing study over time. We find significant evidence of nonlinear associations in both studies and successfully describe these patterns with our method. We give detailed instructions and provide a working program that allows others to directly implement this method in their own analyses.     

An analysis of concanavalin A-mediated agglutination in two Chinese hamster ovary subclones whose surface phenotypes respond to maintenance in medium supplemented with dibutyryl cyclic AMP. V. Biochemical composition of the plasma membrane.

We have used two Chinese hamster ovary subclones whose surface phenotype has been extensively investigated with regard concanavalin A-mediated cell-cell agglutination and concanavalin A-induced receptor site clustering to investigate what changes in membrane composition, if any, can be correlated with the concanavalin A-detected changes in surface phenotype. These cell clones are uniquely disposed for this purpose since maintenance of the cells under different growth conditions produces changes in agglutinability and receptor site mobility in one cell clone (H-7W) but not the other (K-1). After extensive characterization of the surface membranes of these two subclones we have been unable to identify any change in the membrane peptides, glycopeptide, cholesterol, or fatty acid composition which can be directly correlated with the concanavalin A-detected surface phenotypes. It is of particular interest to note that we have been unable to correlate the presence or absence of the large external transformation-sensitive glycoprotein with the relative mobility of the lectin receptors or with the degree of concanavalin A-mediated cell agglutination. Furthermore we have been unable, in this system, to corroborate earlier data suggesting a role for cholesterol in determining the relative mobility of the lectin receptors. Thus using a cell system consisting of genetically matched cell clones, we have been unable to identify any changes in the biochemical composition of the plasma membrane which might be associated with the surface phenotypes detected by concanavalin A.     

Family-based tests of association and linkage that use unaffected sibs, covariates, and interactions

We extend the methodology for family-based tests of association and linkage to allow for both variation in the phenotypes of subjects and incorporation of covariates into general-score tests of association. We use standard association models for a phenotype and any number of predictors. We then construct a score statistic, using likelihoods for the distribution of phenotype, given genotype. The distribution of the score is computed as a function of offspring genotypes, conditional on parental genotypes and trait values for offspring and parents. This approach provides a natural extension of the transmission/disequilibrium test to any phenotype and to multiple genes or environmental factors and allows the study of gene-gene and gene-environment interaction. When the trait varies among subjects or when covariates are included in the association model, the score statistic depends on one or more nuisance parameters. We suggest two approaches for obtaining parameter estimates: (1) choosing the estimate that minimizes the variance of the test statistic and (2) maximizing the statistic over a nuisance parameter and using a corrected P value. We apply our methods to a sample of families with attention-deficit/hyperactivity disorder and provide examples of how covariates and gene-environment and gene-gene interactions can be incorporated.     

A powerful adaptive microbiome-based association test for microbial association signals with diverse sparsity levels

The dysbiosis of microbiome may have negative effects on a host phenotype. The microbes related to the host phenotype are regarded as microbial association signals. Recently, statistical methods based on microbiome-phenotype association tests have been extensively developed to detect these association signals. However, the currently available methods do not perform well to detect microbial association signals when dealing with diverse sparsity levels (i.e., sparse, low sparse, non-sparse). Actually, the real association patterns related to different host phenotypes are not unique. Here, we propose a powerful and adaptive microbiome-based association test to detect microbial association signals with diverse sparsity levels, designated as MiATDS. In particular, we define probability degree to measure the associations between microbes and the host phenotype and introduce the adaptive weighted sum of powered score tests by considering both probability degree and phylogenetic information. We design numerous simulation experiments for the task of detecting association signals with diverse sparsity levels to prove the performance of the method. We find that type I error rates can be well-controlled and MiATDS shows superior efficiency on the power. By applying to real data analysis, MiATDS displays reliable practicability too. The R package is available at https://github.com/XiaoyunHuang33/MiATDS.     

Growth consequences of plasticity of plant traits in response to light conditions

Summary We present a method for quantifying the growth advantage, if any, that results from the plasticity of plant traits in response to growth in high vs. low resource levels. The method, which uses two phenotypes and two resource levels, quantifies the average advantage that a phenotype has, in its own set of conditions, over the other phenotype. The method is applied to the growth of two phenotypes of Abutilon theophrasti, induced by high and low light intensity, in response to two levels of incident light intensity. We calculated the growth advantage first using relative growth rate, and second using whole-plant photosynthetic assimilation rate, as the response variable. Then we used the photosynthetic responses to changes in light intensity to calculate changes in growth rates of each phenotype when exposed to a change in light conditions. These three quantifications of growth advantage broadly agree with one another. Despite the great plasticity of its traits induced by growth in high vs. low light intensity, whole-plant plasticity did not allow Abutilon theophrasti to exhibit a significant growth advantage under these conditions. Indeed, the relative growth rate of the low light phenotype greatly exceeded that of the high light phenotype in high incident light conditions. This may have resulted from the higher leaf area ratio of the low light phenotype. Furthermore, the high light phenotype had significantly greater transpiration rate in both light conditions. For these reasons we suggest that light-induced plasticity of traits in Abutilon theophrasti may confer advantage in response to the variation in vapor pressure deficit that is associated with variation in light intensity. Light-induced plasticity may also be advantageous because under high incident light conditions the high-light phenotype has greater reproductive allocation than the low-light phenotype.     

Crown condition assessment: An accurate,precise and efficient method with broad applicability to Eucalyptus

Assessment of crown condition is a useful tool for monitoring forest health and is used widely to assess tree dieback and decline. Multiple methods used to assess eucalypt crown condition are difficult to compare across studies. Furthermore, the relative effectiveness of the available methods has not been evaluated. The objective of this study was to find an accurate, precise and efficient method for the assessment of crown condition in eucalypts. Four widely used methods were used to assess the crown condition of 516 eucalypt trees from Tasmania and Western Australia. Two of the methods assessed individual crown condition parameters (single‐parameter methods) which could then be added to give an overall score for condition (additive parameter). The other two methods use a single score to encompass many crown parameters (combined‐parameter methods). A selection of trees was scored on multiple occasions and by multiple assessors to determine repeatability and reproducibility. Data were analysed by Spearman's correlations and principal components analysis. All scored parameters were positively correlated to varying degrees. All parameters were distributed along a single‐principal components analysis axis, with the parameters from the single‐parameter methods having the greatest weightings. In order to address the objective of this study five criteria were developed for consideration of parameters that assess crown condition: (i) capacity to assess dieback (high correlation with other crown parameters, and capacity to indicate the presence of dieback symptoms in the crown); (ii) observer bias (sensitivity to minor change and small difference between observers); (iii) repeatability (sensitivity to minor change and small difference between years); (iv) capacity to assess different species; and (v) efficiency to score. Methods that best met these criteria used additive parameters derived from the crown parameters of primary crown dieback, epicormic growth and either crown shrinkage or dead branches. The single most useful parameter for assessment of eucalypt crown condition was primary crown dieback. This parameter was found to be the most accurate and precise measure of crown condition and is efficient to score. Primary crown dieback is recommended as the standard method for assessment of crown condition of eucalypt trees.     

Phenotypic switching in Cryptococcus neoformans

Cryptococcus neoformans strains exhibit considerable phenotype variability with regards to the capsular polysaccharide, sterol composition of the cell wall, and cell and colony morphology. Phenotypic changes can occur spontaneously during in vitro passage of strains or during chronic infection in vivo and may be associated with differences in virulence. Studies from our laboratory have demonstrated that phenotype variability can be the result of phenotypic switching. Phenotypic switching is defined as a reversible change of an observable colony phenotype that occurs at a frequency above the expected frequency for somatic mutations. This implies that phenotypic switching represents controlled and programmed changes in this pathogenic yeast rather than random mutations. We have shown that a phenotypic switch from a smooth colony phenotype to a mucoid colony phenotype occurs in vitro and in vivo during chronic infection of mice. More importantly we have now demonstrated that the switch is associated with an increase in virulence and a change in the host immune response. Implications of these findings for the pathogenesis of cryptococcosis are discussed.     

Effects of long-range correlations in DNA on sequence alignment score statistics.

Long-range correlations in genomic base composition are a ubiquitous statistical feature among many eukaryotic genomes. In this article, these correlations are shown to substantially influence the statistics of sequence alignment scores. Using a Gaussian approximation to model the correlated score landscape, we calculate the corrections to the scale parameter lambda of the extreme value distribution of alignment scores. Our approximate analytic results are supported by a detailed numerical study based on a simple algorithm to efficiently generate long-range correlated random sequences. We find both, mean and exponential tail of the score distribution for long-range correlated sequences to be substantially shifted compared to random sequences with independent nucleotides. The significance of measured alignment scores will therefore change upon incorporation of the correlations in the null model. We discuss the magnitude of this effect in a biological context.     

Causes of covariation of phenotypic traits among populations

Morphological and life-history traits often vary among populations of a species. Traits generally do not vary independently, but show patterns of covariation that can arise from genetic and environmental influences on phenotype. Covariance of traits may arise at an among-population level when genetically influenced traits diverge among populations in a correlated manner. Genetic correlations caused by pleiotropy and/or gene linkage can cause traits to evolve together, but among-population covariance can also arise among traits that are not genetically correlated. For example, “selective covariance” can arise when natural selection directly causes correlated change in a suite of traits. Similarly, mutation, migration, and drift may also sometimes cause correlated genetic changes among populations. Because covariation of traits among populations can arise by several different processes, the evolution of suites of traits must be interpreted with great caution. We discuss the sources of among-population covariance and illustrate one approach to identifying the sources' using data on floral traits of Dalechampia scandens (Euphorbiaceae).     

Evolutionary convergence on sleep loss in cavefish populations

Patterns of sleep vary widely among species, but the functional and evolutionary principles responsible for this diversity remain unknown. The characin fish, Astyanax mexicanus, has eyed surface and numerous blind cave populations. The cave populations are largely independent in their origins, and the species is ideal for studying the genetic bases of convergent evolution. Here we show that this system is also uniquely valuable for the investigation of variability in patterns of sleep. We find that a clearly defined change in ecological conditions, from surface to cave, is correlated with a dramatic reduction in sleep in three independently derived cave populations of A. mexicanus. Analyses of surface × cave hybrids show that the alleles for reduced sleep in the Pachón and Tinaja cave populations are dominant in effect to the surface alleles. Genetic analysis of hybrids between surface and Pachón cavefish suggests that only a small number of loci with dominant effects are involved. Our results demonstrate that sleep is an evolutionarily labile phenotype, highly responsive to changes in ecological conditions. To our knowledge, this is the first example of a single species with a convergence on sleep loss exhibited by several independently evolved populations correlated with population-specific ecologies.     

Conversion from electrocardiosignals to equivalent electrical sources on heart surface

In biomarker discovery, applying domain knowledge is an effective approach to eliminating false positive features, prioritizing functionally impactful markers and facilitating the interpretation of predictive signatures. Several computational methods have been developed that formulate the knowledge-based biomarker discovery as a feature selection problem guided by prior information. These methods often require that prior information is encoded as a single score and the algorithms are optimized for biological knowledge of a specific type. However, in practice, domain knowledge from diverse resources can provide complementary information. But no current methods can integrate heterogeneous prior information for biomarker discovery. To address this problem, we developed the Know-GRRF (know-guided regularized random forest) method that enables dynamic incorporation of domain knowledge from multiple disciplines to guide feature selection. Know-GRRF embeds domain knowledge in a regularized random forest framework. It combines prior information from multiple domains in a linear model to derive a composite score, which, together with other tuning parameters, controls the regularization of the random forests model. Know-GRRF concurrently optimizes the weight given to each type of domain knowledge and other tuning parameters to minimize the AIC of out-of-bag predictions. The objective is to select a compact feature subset that has a high discriminative power and strong functional relevance to the biological phenotype. Via rigorous simulations, we show that Know-GRRF guided by multiple-domain prior information outperforms feature selection methods guided by single-domain prior information or no prior information. We then applied Known-GRRF to a real-world study to identify prognostic biomarkers of prostate cancers. We evaluated the combination of cancer-related gene annotations, evolutionary conservation and pre-computed statistical scores as the prior knowledge to assemble a panel of biomarkers. We discovered a compact set of biomarkers with significant improvements on prediction accuracies. Know-GRRF is a powerful novel method to incorporate knowledge from multiple domains for feature selection. It has a broad range of applications in biomarker discoveries. We implemented this method and released a KnowGRRF package in the R/CRAN archive.     

Identification and analysis of deleterious human SNPs

We have developed two methods of identifying which non-synonomous single base changes have a deleterious effect on protein function in vivo. One method, described elsewhere, analyzes the effect of the resulting amino acid change on protein stability, utilizing structural information. The other method, introduced here, makes use of the conservation and type of residues observed at a base change position within a protein family. A machine learning technique, the support vector machine, is trained on single amino acid changes that cause monogenic disease, with a control set of amino acid changes fixed between species. Both methods are used to identify deleterious single nucleotide polymorphisms (SNPs) in the human population. After carefully controlling for errors, we find that approximately one quarter of known non-synonymous SNPs are deleterious by these criteria, providing a set of possible contributors to human complex disease traits.