A machine learning approach for genome-wide prediction of morbid and druggable human genes based on systems-level data

Background: The genome-wide identification of both morbid genes, i.e., those genes whose mutations cause hereditary human diseases, and druggable genes, i.e., genes coding for proteins whose modulation by small molecules elicits phenotypic effects, requires experimental approaches that are time-consuming and laborious. Thus, a computational approach which could accurately predict such genes on a genome-wide scale would be invaluable for accelerating the pace of discovery of causal relationships between genes and diseases as well as the determination of druggability of gene products.Results: In this paper we propose a machine learning-based computational approach to predict morbid and druggable genes on a genome-wide scale. For this purpose, we constructed a decision tree-based meta-classifier and trained it on datasets containing, for each morbid and druggable gene, network topological features, tissue expression profile and subcellular localization data as learning attributes. This meta-classifier correctly recovered 65% of known morbid genes with a precision of 66% and correctly recovered 78% of known druggable genes with a precision of 75%. It was than used to assign morbidity and druggability scores to genes not known to be morbid and druggable and we showed a good match between these scores and literature data. Finally, we generated decision trees by training the J48 algorithm on the morbidity and druggability datasets to discover cellular rules for morbidity and druggability and, among the rules, we found that the number of regulating transcription factors and plasma membrane localization are the most important factors to morbidity and druggability, respectively.Conclusions: We were able to demonstrate that network topological features along with tissue expression profile and subcellular localization can reliably predict human morbid and druggable genes on a genome-wide scale. Moreover, by constructing decision trees based on these data, we could discover cellular rules governing morbidity and druggability.

Baixas Doses de 5-fluorouracil aumentam a atividade antitumoral de células dendríticas transfectadas com RNA de células de câncer colorretal

Pós-graduação em Patologia - FMB

Identificação de genes candidatos à indução do florescimento em cana-de-açúcar em câmara de fotoperíodo

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Bioinformatic methods in applied genomic research

Here I will focus on three main topics that best address and include the projects I have been working in during my three year PhD period that I have spent in different research laboratories addressing both computationally and practically important problems all related to modern molecular genomics. The first topic is the use of livestock species (pigs) as a model of obesity, a complex human dysfunction. My efforts here concern the detection and annotation of Single Nucleotide Polymorphisms. I developed a pipeline for mining human and porcine sequences. Starting from a set of human genes related with obesity the platform returns a list of annotated porcine SNPs extracted from a new set of potential obesity-genes. 565 of these SNPs were analyzed on an Illumina chip to test the involvement in obesity on a population composed by more than 500 pigs. Results will be discussed. All the computational analysis and experiments were done in collaboration with the Biocomputing group and Dr.Luca Fontanesi, respectively, under the direction of prof. Rita Casadio at the Bologna University, Italy. The second topic concerns developing a methodology, based on Factor Analysis, to simultaneously mine information from different levels of biological organization. With specific test cases we develop models of the complexity of the mRNA-miRNA molecular interaction in brain tumors measured indirectly by microarray and quantitative PCR. This work was done under the supervision of Prof. Christine Nardini, at the “CAS-MPG Partner Institute for Computational Biology” of Shangai, China (co-founded by the Max Planck Society and the Chinese Academy of Sciences jointly) The third topic concerns the development of a new method to overcome the variety of PCR technologies routinely adopted to characterize unknown flanking DNA regions of a viral integration locus of the human genome after clinical gene therapy. This new method is entirely based on next generation sequencing and it reduces the time required to detect insertion sites, decreasing the complexity of the procedure. This work was done in collaboration with the group of Dr. Manfred Schmidt at the Nationales Centrum für Tumorerkrankungen (Heidelberg, Germany) supervised by Dr. Annette Deichmann and Dr. Ali Nowrouzi. Furthermore I add as an Appendix the description of a R package for gene network reconstruction that I helped to develop for scientific usage (http://www.bioconductor.org/help/bioc-views/release/bioc/html/BUS.html).

Computation of direction selectivity in retinal starburst amacrine cell dendrites – studied using electrophysiological recordings and two-photon imaging

Neuronal circuits in the retina analyze images according to qualitative aspects such as color or motion, before the information is transmitted to higher visual areas of the brain. One example, studied for over the last four decades, is the detection of motion direction in ‘direction selective’ neurons. Recently, the starburst amacrine cell, one type of retinal interneuron, has emerged as an essential player in the computation of direction selectivity. In this study the mechanisms underlying the computation of direction selective calcium signals in starburst cell dendrites were investigated using whole-cell electrical recordings and two-photon calcium imaging. Analysis of the somatic electrical responses to visual stimulation and pharmacological agents indicated that the directional signal (i) is not computed presynaptically to starburst cells or by inhibitory network interactions. It is thus computed via a cell-intrinsic mechanism, which (ii) depends upon the differential, i.e. direction selective, activation of voltage-gated channels. Optically measuring dendritic calcium signals as a function of somatic voltage suggests (iii) a difference in resting membrane potential between the starburst cell’s soma and its distal dendrites. In conclusion, it is proposed that the mechanism underlying direction selectivity in starburst cell dendrites relies on intrinsic properties of the cell, particularly on the interaction of spatio-temporally structured synaptic inputs with voltage-gated channels, and their differential activation due to a somato-dendritic difference in membrane potential.

Self-sufficient control of urate homeostasis in mice by a synthetic circuit

Synthetic biology has shown that the metabolic behavior of mammalian cells can be altered by genetic devices such as epigenetic and hysteretic switches, timers and oscillators, biocomputers, hormone systems and heterologous metabolic shunts. To explore the potential of such devices for therapeutic strategies, we designed a synthetic mammalian circuit to maintain uric acid homeostasis in the bloodstream, disturbance of which is associated with tumor lysis syndrome and gout. This synthetic device consists of a modified Deinococcus radiodurans-derived protein that senses uric acids levels and triggers dose-dependent derepression of a secretion-engineered Aspergillus flavus urate oxidase that eliminates uric acid. In urate oxidase-deficient mice, which develop acute hyperuricemia, the synthetic circuit decreased blood urate concentration to stable sub-pathologic levels in a dose-dependent manner and reduced uric acid crystal deposits in the kidney. Synthetic gene-network devices providing self-sufficient control of pathologic metabolites represent molecular prostheses, which may foster advances in future gene- and cell-based therapies.

STATISTICAL METHODS FOR RARE AND COMMON VARIANT ASSOCIATION STUDIES

Complex human diseases are a major challenge for biological research. The goal of my research is to develop effective methods for biostatistics in order to create more opportunities for the prevention and cure of human diseases. This dissertation proposes statistical technologies that have the ability of being adapted to sequencing data in family-based designs, and that account for joint effects as well as gene-gene and gene-environment interactions in the GWA studies. The framework includes statistical methods for rare and common variant association studies. Although next-generation DNA sequencing technologies have made rare variant association studies feasible, the development of powerful statistical methods for rare variant association studies is still underway. Chapter 2 demonstrates two adaptive weighting methods for rare variant association studies based on family data for quantitative traits. The results show that both proposed methods are robust to population stratification, robust to the direction and magnitude of the effects of causal variants, and more powerful than the methods using weights suggested by Madsen and Browning [2009]. In Chapter 3, I extended the previously proposed test for Testing the effect of an Optimally Weighted combination of variants (TOW) [Sha et al., 2012] for unrelated individuals to TOW &ndash F, TOW for Family &ndash based design. Simulation results show that TOW &ndash F can control for population stratification in wide range of population structures including spatially structured populations, is robust to the directions of effect of causal variants, and is relatively robust to percentage of neutral variants. In GWA studies, this dissertation consists of a two &ndash locus joint effect analysis and a two-stage approach accounting for gene &ndash gene and gene &ndash environment interaction. Chapter 4 proposes a novel two &ndash stage approach, which is promising to identify joint effects, especially for monotonic models. The proposed approach outperforms a single &ndash marker method and a regular two &ndash stage analysis based on the two &ndash locus genotypic test. In Chapter 5, I proposed a gene &ndash based two &ndash stage approach to identify gene &ndash gene and gene &ndash environment interactions in GWA studies which can include rare variants. The two &ndash stage approach is applied to the GAW 17 dataset to identify the interaction between KDR gene and smoking status.

Hebbian analysis of the transformation of medial entorhinal grid-cell inputs to hippocampal place fields.

The discovery of grid cells in the medial entorhinal cortex (MEC) permits the characterization of hippocampal computation in much greater detail than previously possible. The present study addresses how an integrate-and-fire unit driven by grid-cell spike trains may transform the multipeaked, spatial firing pattern of grid cells into the single-peaked activity that is typical of hippocampal place cells. Previous studies have shown that in the absence of network interactions, this transformation can succeed only if the place cell receives inputs from grids with overlapping vertices at the location of the place cell's firing field. In our simulations, the selection of these inputs was accomplished by fast Hebbian plasticity alone. The resulting nonlinear process was acutely sensitive to small input variations. Simulations differing only in the exact spike timing of grid cells produced different field locations for the same place cells. Place fields became concentrated in areas that correlated with the initial trajectory of the animal; the introduction of feedback inhibitory cells reduced this bias. These results suggest distinct roles for plasticity of the perforant path synapses and for competition via feedback inhibition in the formation of place fields in a novel environment. Furthermore, they imply that variability in MEC spiking patterns or in the rat's trajectory is sufficient for generating a distinct population code in a novel environment and suggest that recalling this code in a familiar environment involves additional inputs and/or a different mode of operation of the network.

A conserved role for p48 homologs in protecting dopaminergic neurons from oxidative stress.

Parkinson's disease (PD) is the most common neurodegenerative movement disorder characterized by the progressive loss of dopaminergic (DA) neurons. Both environmental and genetic factors are thought to contribute to the pathogenesis of PD. Although several genes linked to rare familial PD have been identified, endogenous risk factors for sporadic PD, which account for the majority of PD cases, remain largely unknown. Genome-wide association studies have identified many single nucleotide polymorphisms associated with sporadic PD in neurodevelopmental genes including the transcription factor p48/ptf1a. Here we investigate whether p48 plays a role in the survival of DA neurons in Drosophila melanogaster and Caenorhabditis elegans. We show that a Drosophila p48 homolog, 48-related-2 (Fer2), is expressed in and required for the development and survival of DA neurons in the protocerebral anterior medial (PAM) cluster. Loss of Fer2 expression in adulthood causes progressive PAM neuron degeneration in aging flies along with mitochondrial dysfunction and elevated reactive oxygen species (ROS) production, leading to the progressive locomotor deficits. The oxidative stress challenge upregulates Fer2 expression and exacerbates the PAM neuron degeneration in Fer2 loss-of-function mutants. hlh-13, the worm homolog of p48, is also expressed in DA neurons. Unlike the fly counterpart, hlh-13 loss-of-function does not impair development or survival of DA neurons under normal growth conditions. Yet, similar to Fer2, hlh-13 expression is upregulated upon an acute oxidative challenge and is required for the survival of DA neurons under oxidative stress in adult worms. Taken together, our results indicate that p48 homologs share a role in protecting DA neurons from oxidative stress and degeneration, and suggest that loss-of-function of p48 homologs in flies and worms provides novel tools to study gene-environmental interactions affecting DA neuron survival.

The effect of genotype-by-environment interaction on longitudinal triglyceride profiles in the Bogalusa Heart Study

Triglyceride levels are a component of plasma lipids that are thought to be an important risk factor for coronary heart disease and are influenced by genetic and environmental factors, such as single nucleotide polymorphisms (SNPs), alcohol intake, and smoking. This study used longitudinal data from the Bogalusa Heart Study, a biracial community-based survey of cardiovascular disease risk factors. A sample of 1191 individuals, 4 to 38 years of age, was measured multiple times from 1973 to 2000. The study sample consisted of 730 white and 461 African American participants. Individual growth models were developed in order to assess gene-environment interactions affecting plasma triglycerides over time. After testing for inclusion of significant covariates and interactions, final models, each accounting for the effects of a different SNP, were assessed for fit and normality. After adjustment for all other covariates and interactions, LIPC -514C/T was found to interact with age3, age2, and age and a non-significant interaction of CETP -971G/A genotype with smoking status was found (p = 0.0812). Ever-smokers had higher triglyceride levels than never smokers, but persons heterozygous at this locus, about half of both races, had higher triglyceride levels after smoking cessation compared to current smokers. Since tobacco products increase free fatty acids circulating in the bloodstream, smoking cessation programs have the potential to ultimately reduce triglyceride levels for many persons. However, due to the effect of smoking cessation on the triglyceride levels of CETP -971G/A heterozygotes, the need for smoking prevention programs is also demonstrated. Both smoking cessation and prevention programs would have a great public health impact on minimizing triglyceride levels and ultimately reducing heart disease. ^

Genetic variations in the PI3K-AKT-mTOR pathway and bladder cancer susceptibility and clinical outcomes

Bladder cancer is the fourth most common cancer in men in the United States. There is compelling evidence supporting that genetic variations contribute to the risk and outcomes of bladder cancer. The PI3K-AKT-mTOR pathway is a major cellular pathway involved in proliferation, invasion, inflammation, tumorigenesis, and drug response. Somatic aberrations of PI3K-AKT-mTOR pathway are frequent events in several cancers including bladder cancer; however, no studies have investigated the role of germline genetic variations in this pathway in bladder cancer. In this project, we used a large case control study to evaluate the associations of a comprehensive catalogue of SNPs in this pathway with bladder cancer risk and outcomes. Three SNPs in RAPTOR were significantly associated with susceptibility: rs11653499 (OR: 1.79, 95%CI: 1.24–2.60), rs7211818 (OR: 2.13, 95%CI: 1.35–3.36), and rs7212142 (OR: 1.57, 95%CI: 1.19–2.07). Two haplotypes constructed from these 3 SNPs were also associated with bladder cancer risk. In combined analysis, a significant trend was observed for increased risk with an increase in the number of unfavorable genotypes (P for trend<0.001). Classification and regression tree analysis identified potential gene-environment interactions between RPS6KA5 rs11653499 and smoking. In superficial bladder cancer, we found that PTEN rs1234219 and rs11202600, TSC1 rs7040593, RAPTOR rs901065, and PIK3R1 rs251404 were significantly associated with recurrence in patients receiving BCG. In muscle invasive and metastatic bladder cancer, AKT2 rs3730050, PIK3R1 rs10515074, and RAPTOR rs9906827 were associated with survival. Survival tree analysis revealed potential gene-gene interactions: patients carrying the unfavorable genotypes of PTEN rs1234219 and TSC1 rs704059 exhibited a 5.24-fold (95% CI: 2.44–11.24) increased risk of recurrence. In combined analysis, with the increasing number of unfavorable genotypes, there was a significant trend of higher risk of recurrence and death (P for trend<0.001) in Cox proportional hazard regression analysis, and shorter event (recurrence and death) free survival in Kaplan-Meier estimates (P log rank<0.001). This study strongly suggests that genetic variations in PI3K-AKT-mTOR pathway play an important role in bladder cancer development. The identified SNPs, if validated in further studies, may become valuable biomarkers in assessing an individual's cancer risk, predicting prognosis and treatment response, and facilitating physicians to make individualized treatment decisions. ^

Detecting genetic and nutritional lung cancer risk factors related to folate metabolism using Bayesian generalized linear models

Complex diseases, such as cancer, are caused by various genetic and environmental factors, and their interactions. Joint analysis of these factors and their interactions would increase the power to detect risk factors but is statistically. Bayesian generalized linear models using student-t prior distributions on coefficients, is a novel method to simultaneously analyze genetic factors, environmental factors, and interactions. I performed simulation studies using three different disease models and demonstrated that the variable selection performance of Bayesian generalized linear models is comparable to that of Bayesian stochastic search variable selection, an improved method for variable selection when compared to standard methods. I further evaluated the variable selection performance of Bayesian generalized linear models using different numbers of candidate covariates and different sample sizes, and provided a guideline for required sample size to achieve a high power of variable selection using Bayesian generalize linear models, considering different scales of number of candidate covariates. ^ Polymorphisms in folate metabolism genes and nutritional factors have been previously associated with lung cancer risk. In this study, I simultaneously analyzed 115 tag SNPs in folate metabolism genes, 14 nutritional factors, and all possible genetic-nutritional interactions from 1239 lung cancer cases and 1692 controls using Bayesian generalized linear models stratified by never, former, and current smoking status. SNPs in MTRR were significantly associated with lung cancer risk across never, former, and current smokers. In never smokers, three SNPs in TYMS and three gene-nutrient interactions, including an interaction between SHMT1 and vitamin B12, an interaction between MTRR and total fat intake, and an interaction between MTR and alcohol use, were also identified as associated with lung cancer risk. These lung cancer risk factors are worthy of further investigation.^

Collective Organization of Complex Networks: Emergent Scales, Vulnerability, and Targeting of Desired Dynamics

Cuando una colectividad de sistemas dinámicos acoplados mediante una estructura irregular de interacciones evoluciona, se observan dinámicas de gran complejidad y fenómenos emergentes imposibles de predecir a partir de las propiedades de los sistemas individuales. El objetivo principal de esta tesis es precisamente avanzar en nuestra comprensión de la relación existente entre la topología de interacciones y las dinámicas colectivas que una red compleja es capaz de mantener. Siendo este un tema amplio que se puede abordar desde distintos puntos de vista, en esta tesis se han estudiado tres problemas importantes dentro del mismo que están relacionados entre sí. Por un lado, en numerosos sistemas naturales y artificiales que se pueden describir mediante una red compleja la topología no es estática, sino que depende de la dinámica que se desarrolla en la red: un ejemplo son las redes de neuronas del cerebro. En estas redes adaptativas la propia topología emerge como consecuencia de una autoorganización del sistema. Para conocer mejor cómo pueden emerger espontáneamente las propiedades comúnmente observadas en redes reales, hemos estudiado el comportamiento de sistemas que evolucionan según reglas adaptativas locales con base empírica. Nuestros resultados numéricos y analíticos muestran que la autoorganización del sistema da lugar a dos de las propiedades más universales de las redes complejas: a escala mesoscópica, la aparición de una estructura de comunidades, y, a escala macroscópica, la existencia de una ley de potencias en la distribución de las interacciones en la red. El hecho de que estas propiedades aparecen en dos modelos con leyes de evolución cuantitativamente distintas que siguen unos mismos principios adaptativos sugiere que estamos ante un fenómeno que puede ser muy general, y estar en el origen de estas propiedades en sistemas reales. En segundo lugar, proponemos una medida que permite clasificar los elementos de una red compleja en función de su relevancia para el mantenimiento de dinámicas colectivas. En concreto, estudiamos la vulnerabilidad de los distintos elementos de una red frente a perturbaciones o grandes fluctuaciones, entendida como una medida del impacto que estos acontecimientos externos tienen en la interrupción de una dinámica colectiva. Los resultados que se obtienen indican que la vulnerabilidad dinámica es sobre todo dependiente de propiedades locales, por tanto nuestras conclusiones abarcan diferentes topologías, y muestran la existencia de una dependencia no trivial entre la vulnerabilidad y la conectividad de los elementos de una red. Finalmente, proponemos una estrategia de imposición de una dinámica objetivo genérica en una red dada e investigamos su validez en redes con diversas topologías que mantienen regímenes dinámicos turbulentos. Se obtiene como resultado que las redes heterogéneas (y la amplia mayora de las redes reales estudiadas lo son) son las más adecuadas para nuestra estrategia de targeting de dinámicas deseadas, siendo la estrategia muy efectiva incluso en caso de disponer de un conocimiento muy imperfecto de la topología de la red. Aparte de la relevancia teórica para la comprensión de fenómenos colectivos en sistemas complejos, los métodos y resultados propuestos podrán dar lugar a aplicaciones en sistemas experimentales y tecnológicos, como por ejemplo los sistemas neuronales in vitro, el sistema nervioso central (en el estudio de actividades síncronas de carácter patológico), las redes eléctricas o los sistemas de comunicaciones. ABSTRACT The time evolution of an ensemble of dynamical systems coupled through an irregular interaction scheme gives rise to dynamics of great of complexity and emergent phenomena that cannot be predicted from the properties of the individual systems. The main objective of this thesis is precisely to increase our understanding of the interplay between the interaction topology and the collective dynamics that a complex network can support. This is a very broad subject, so in this thesis we will limit ourselves to the study of three relevant problems that have strong connections among them. First, it is a well-known fact that in many natural and manmade systems that can be represented as complex networks the topology is not static; rather, it depends on the dynamics taking place on the network (as it happens, for instance, in the neuronal networks in the brain). In these adaptive networks the topology itself emerges from the self-organization in the system. To better understand how the properties that are commonly observed in real networks spontaneously emerge, we have studied the behavior of systems that evolve according to local adaptive rules that are empirically motivated. Our numerical and analytical results show that self-organization brings about two of the most universally found properties in complex networks: at the mesoscopic scale, the appearance of a community structure, and, at the macroscopic scale, the existence of a power law in the weight distribution of the network interactions. The fact that these properties show up in two models with quantitatively different mechanisms that follow the same general adaptive principles suggests that our results may be generalized to other systems as well, and they may be behind the origin of these properties in some real systems. We also propose a new measure that provides a ranking of the elements in a network in terms of their relevance for the maintenance of collective dynamics. Specifically, we study the vulnerability of the elements under perturbations or large fluctuations, interpreted as a measure of the impact these external events have on the disruption of collective motion. Our results suggest that the dynamic vulnerability measure depends largely on local properties (our conclusions thus being valid for different topologies) and they show a non-trivial dependence of the vulnerability on the connectivity of the network elements. Finally, we propose a strategy for the imposition of generic goal dynamics on a given network, and we explore its performance in networks with different topologies that support turbulent dynamical regimes. It turns out that heterogeneous networks (and most real networks that have been studied belong in this category) are the most suitable for our strategy for the targeting of desired dynamics, the strategy being very effective even when the knowledge on the network topology is far from accurate. Aside from their theoretical relevance for the understanding of collective phenomena in complex systems, the methods and results here discussed might lead to applications in experimental and technological systems, such as in vitro neuronal systems, the central nervous system (where pathological synchronous activity sometimes occurs), communication systems or power grids.

Targeted expression of teashirt induces ectopic eyes in Drosophila

teashirt was initially identified as a gene required for the specification of the trunk segments in Drosophila embryogenesis and encodes a transcription factor with zinc finger motifs. We report here that targeted expression of teashirt in imaginal discs is sufficient to induce ectopic eye formation in non-eye tissues, a phenotype similar to that produced from targeted expression of eyeless, dachshund, and eyes absent. Furthermore, teashirt and eyeless induce the expression of each other, suggesting that teashirt is part of the gene network that functions to specify eye identity.

The nuclear matrix protein NMP-1 is the transcription factor YY1.

NMP-1 was initially identified as a nuclear matrix-associated DNA-binding factor that exhibits sequence-specific recognition for the site IV regulatory element of a histone H4 gene. This distal promoter domain is a nuclear matrix interaction site. In the present study, we show that NMP-1 is the multifunctional transcription factor YY1. Gel-shift and Western blot analyses demonstrate that NMP-1 is immunoreactive with YY1 antibody. Furthermore, purified YY1 protein specifically recognizes site IV and reconstitutes the NMP-1 complex. Western blot and gel-shift analyses indicate that YY1 is present within the nuclear matrix. In situ immunofluorescence studies show that a significant fraction of YY1 is localized in the nuclear matrix, principally but not exclusively associated with residual nucleoli. Our results confirm that NMP-1/YY1 is a ubiquitous protein that is present in both human cells and in rat osteosarcoma ROS 17/2.8 cells. The finding that NMP-1 is identical to YY1 suggests that this transcriptional regulator may mediate gene-matrix interactions. Our results are consistent with the concept that the nuclear matrix may functionally compartmentalize the eukaryotic nucleus to support regulation of gene expression.