Functional genomic hypothesis generation and experimentation by a robot scientist

King R. D., Whelan, K. E., Jones, F. M., Reiser, P. G. K., Bryant, C. H., Muggleton, S., Kell, D. B. and Oliver, S. G. (2004) Functional genomic hypothesis generation and experimentation by a robot scientist. Nature 427 (6971) p247-252

Clinical reasoning: The relative contribution of identification, interpretation and hypothesis errors to misdiagnosis

The aim of this study was to identify and describe the types of errors in clinical reasoning that contribute to poor diagnostic performance at different levels of medical training and experience. Three cohorts of subjects, second- and fourth- (final) year medical students and a group of general practitioners, completed a set of clinical reasoning problems. The responses of those whose scores fell below the 25th centile were analysed to establish the stage of the clinical reasoning process - identification of relevant information, interpretation or hypothesis generation - at which most errors occurred and whether this was dependent on problem difficulty and level of medical experience. Results indicate that hypothesis errors decrease as expertise increases but that identification and interpretation errors increase. This may be due to inappropriate use of pattern recognition or to failure of the knowledge base. Furthermore, although hypothesis errors increased in line with problem difficulty, identification and interpretation errors decreased. A possible explanation is that as problem difficulty increases, subjects at all levels of expertise are less able to differentiate between relevant and irrelevant clinical features and so give equal consideration to all information contained within a case. It is concluded that the development of clinical reasoning in medical students throughout the course of their pre-clinical and clinical education may be enhanced by both an analysis of the clinical reasoning process and a specific focus on each of the stages at which errors commonly occur.

Quantitative assessment of urea generation and elimination in healthy dogs and in dogs with chronic kidney disease

BACKGROUND: Kinetic assessment of urea, the main end product of protein metabolism, could serve to assess protein catabolism in dogs with chronic kidney disease (CKD). Protein malnutrition and catabolism are poorly documented in CKD and they often are neglected clinically because of a lack of appropriate evaluation tools. HYPOTHESIS: Generation and excretion of urea are altered in dogs with CKD. ANIMALS: Nine dogs with spontaneous CKD (IRIS stages 2-4) and 5 healthy research dogs. METHODS: Endogenous renal clearance (Clrenal) of urea and creatinine was measured first. Exogenous plasma clearance (Clplasma, total body clearance) of the 2 markers then was determined by an IV infusion of urea (250-1,000 mg/kg over 20 minutes) and an IV bolus of creatinine (40 mg/kg). Extrarenal clearance (Clextra) was defined as the difference between Clplasma)and Clrenal. Endogenous urea generation was computed assuming steady-state conditions. RESULTS: Median Clrenal and Clextra of urea were 2.17 and 0.21 mL/min/kg in healthy dogs and 0.37 and 0.28 mL/min/kg in CKD dogs. The proportion of urea cleared by extrarenal route was markedly higher in dogs with glomerular filtration rate<1 mL/kg/min than in normal dogs, reaching up to 85% of the total clearance. A comparable pattern was observed for creatinine excretion, except in 1 dog, Clextra remained<20% of Clplasma. CONCLUSION: Extrarenal pathways of urea excretion are predominant in dogs with advanced CKD and justify exploring adjunctive therapies based on enteric nitrogen excretion in dogs. A trend toward increased urea generation may indicate increased catabolism in advanced CKD.

Molecular Modelling of Estrogen-Producing Enzymes CYP450 Aromatase and 17beta-Hydroxysteroid Dehydrogenase Type 1

Breast cancer is the most common cancer in women in the western countries. Approximately two-thirds of breast cancer tumours are hormone dependent, requiring estrogens to grow. Estrogens are formed in the human body via a multistep route starting from cholesterol. The final steps in the biosynthesis include the CYP450 aromatase enzyme, converting the male hormones androgens (preferred substrate androstenedione ASD) into estrogens(estrone E1), and the 17beta-HSD1 enzyme, converting the biologically less active E1 into the active hormone 17beta-hydroxyestradiol E2. E2 is bound to the nuclear estrogen receptors causing a cascade of biochemical reactions leading to cell proliferation in normal tissue, and to tumour growth in cancer tissue. Aromatase and 17beta-HSD1 are expressed in or near the breast tumour, locally providing the tissue with estrogens. One approach in treating hormone dependent breast tumours is to block the local estrogen production by inhibiting these two enzymes. Aromatase inhibitors are already on the market in treating breast cancer, despite the lack of an experimentally solved structure. The structure of 17beta-HSD1, on the other hand, has been solved, but no commercial drugs have emerged from the drug discovery projects reported in the literature. Computer-assisted molecular modelling is an invaluable tool in modern drug design projects. Modelling techniques can be used to generate a model of the target protein and to design novel inhibitors for them even if the target protein structure is unknown. Molecular modelling has applications in predicting the activities of theoretical inhibitors and in finding possible active inhibitors from a compound database. Inhibitor binding at atomic level can also be studied with molecular modelling. To clarify the interactions between the aromatase enzyme and its substrate and inhibitors, we generated a homology model based on a mammalian CYP450 enzyme, rabbit progesterone 21-hydroxylase CYP2C5. The model was carefully validated using molecular dynamics simulations (MDS) with and without the natural substrate ASD. Binding orientation of the inhibitors was based on the hypothesis that the inhibitors coordinate to the heme iron, and were studied using MDS. The inhibitors were dietary phytoestrogens, which have been shown to reduce the risk for breast cancer. To further validate the model, the interactions of a commercial breast cancer drug were studied with MDS and ligand–protein docking. In the case of 17beta-HSD1, a 3D QSAR model was generated on the basis of MDS of an enzyme complex with active inhibitor and ligand–protein docking, employing a compound library synthesised in our laboratory. Furthermore, four pharmacophore hypotheses with and without a bound substrate or an inhibitor were developed and used in screening a commercial database of drug-like compounds. The homology model of aromatase showed stable behaviour in MDS and was capable of explaining most of the results from mutagenesis studies. We were able to identify the active site residues contributing to the inhibitor binding, and explain differences in coordination geometry corresponding to the inhibitory activity. Interactions between the inhibitors and aromatase were in agreement with the mutagenesis studies reported for aromatase. Simulations of 17beta-HSD1 with inhibitors revealed an inhibitor binding mode with hydrogen bond interactions previously not reported, and a hydrophobic pocket capable of accommodating a bulky side chain. Pharmacophore hypothesis generation, followed by virtual screening, was able to identify several compounds that can be used in lead compound generation. The visualisation of the interaction fields from the QSAR model and the pharmacophores provided us with novel ideas for inhibitor development in our drug discovery project.

Pharmacogenomics in early-phase clinical development.

Pharmacogenomics (PGx) offers the promise of utilizing genetic fingerprints to predict individual responses to drugs in terms of safety, efficacy and pharmacokinetics. Early-phase clinical trial PGx applications can identify human genome variations that are meaningful to study design, selection of participants, allocation of resources and clinical research ethics. Results can inform later-phase study design and pipeline developmental decisions. Nevertheless, our review of the clinicaltrials.gov database demonstrates that PGx is rarely used by drug developers. Of the total 323 trials that included PGx as an outcome, 80% have been conducted by academic institutions after initial regulatory approval. Barriers for the application of PGx are discussed. We propose a framework for the role of PGx in early-phase drug development and recommend PGx be universally considered in study design, result interpretation and hypothesis generation for later-phase studies, but PGx results from underpowered studies should not be used by themselves to terminate drug-development programs.

Mining biomedical information from scientific literature

The rapid evolution and proliferation of a world-wide computerized network, the Internet, resulted in an overwhelming and constantly growing amount of publicly available data and information, a fact that was also verified in biomedicine. However, the lack of structure of textual data inhibits its direct processing by computational solutions. Information extraction is the task of text mining that intends to automatically collect information from unstructured text data sources. The goal of the work described in this thesis was to build innovative solutions for biomedical information extraction from scientific literature, through the development of simple software artifacts for developers and biocurators, delivering more accurate, usable and faster results. We started by tackling named entity recognition - a crucial initial task - with the development of Gimli, a machine-learning-based solution that follows an incremental approach to optimize extracted linguistic characteristics for each concept type. Afterwards, Totum was built to harmonize concept names provided by heterogeneous systems, delivering a robust solution with improved performance results. Such approach takes advantage of heterogenous corpora to deliver cross-corpus harmonization that is not constrained to specific characteristics. Since previous solutions do not provide links to knowledge bases, Neji was built to streamline the development of complex and custom solutions for biomedical concept name recognition and normalization. This was achieved through a modular and flexible framework focused on speed and performance, integrating a large amount of processing modules optimized for the biomedical domain. To offer on-demand heterogenous biomedical concept identification, we developed BeCAS, a web application, service and widget. We also tackled relation mining by developing TrigNER, a machine-learning-based solution for biomedical event trigger recognition, which applies an automatic algorithm to obtain the best linguistic features and model parameters for each event type. Finally, in order to assist biocurators, Egas was developed to support rapid, interactive and real-time collaborative curation of biomedical documents, through manual and automatic in-line annotation of concepts and relations. Overall, the research work presented in this thesis contributed to a more accurate update of current biomedical knowledge bases, towards improved hypothesis generation and knowledge discovery.

Une activité d'élaboration d'hypothèses pour soutenir le développement du RCI d'étudiantes en sciences infirmières

L'enseignement du raisonnement clinique infirmier (RCI) est une préoccupation importante des formateurs en sciences infirmières depuis plusieurs années. Les étudiantes en sciences infirmières éprouvent des difficultés à formuler des hypothèses cliniques, à savoir trouver les explications pouvant justifier la coexistence d'une combinaison de données cliniques. Pourtant, la formulation d’hypothèses constitue une étape déterminante du RCI. Dans cette étude qualitative exploratoire, nous avons mis à l'essai une activité d'apprentissage par vignette clinique courte (AVCC) qui fournit aux étudiantes l'occasion d'exercer spécifiquement la formulation d'hypothèses cliniques. L'étude visait à documenter la capacité d'étudiantes de troisième année au baccalauréat en sciences infirmières à formuler des hypothèses cliniques durant l'activité. Dix-sept étudiantes ont été recrutées par convenance et divisées en groupes selon leurs disponibilités. Au total, quatre séances ont eu lieu. Les participantes étaient invitées à réfléchir à une vignette clinique courte et à construire un algorithme qui incluait: 1) leurs hypothèses concernant la nature du problème clinique, 2) les éléments d'informations essentiels à rechercher pour vérifier chaque hypothèse et 3) les moyens pour trouver ces informations. L'observation participante, l'enregistrement audio-vidéo et un questionnaire auto-administré ont servi à collecter les données. Les stratégies de RCI décrites par Fonteyn (1998) ont servi de cadre théorique pour guider l’analyse, sous forme de matrices comprenant des verbatims et des notes de terrain. Les résultats suggèrent que l'AVCC stimule la formulation d'hypothèses cliniques et la réactivation des connaissances antérieures. Cette activité pourrait donc être utile en complément d'autres activités éducatives pour favoriser le développement du RCI chez les étudiantes en sciences infirmières.

Natural Object Categorization

This thesis addresses the problem of categorizing natural objects. To provide a criteria for categorization we propose that the purpose of a categorization is to support the inference of unobserved properties of objects from the observed properties. Because no such set of categories can be constructed in an arbitrary world, we present the Principle of Natural Modes as a claim about the structure of the world. We first define an evaluation function that measures how well a set of categories supports the inference goals of the observer. Entropy measures for property uncertainty and category uncertainty are combined through a free parameter that reflects the goals of the observer. Natural categorizations are shown to be those that are stable with respect to this free parameter. The evaluation function is tested in the domain of leaves and is found to be sensitive to the structure of the natural categories corresponding to the different species. We next develop a categorization paradigm that utilizes the categorization evaluation function in recovering natural categories. A statistical hypothesis generation algorithm is presented that is shown to be an effective categorization procedure. Examples drawn from several natural domains are presented, including data known to be a difficult test case for numerical categorization techniques. We next extend the categorization paradigm such that multiple levels of natural categories are recovered; by means of recursively invoking the categorization procedure both the genera and species are recovered in a population of anaerobic bacteria. Finally, a method is presented for evaluating the utility of features in recovering natural categories. This method also provides a mechanism for determining which features are constrained by the different processes present in a multiple modal world.

A CAD-based computer vision system

A vision system for recognizing rigid and articulated three-dimensional objects in two-dimensional images is described. Geometrical models are extracted from a commercial computer aided design package. The models are then augmented with appearance and functional information which improves the system's hypothesis generation, hypothesis verification, and pose refinement. Significant advantages over existing CAD-based vision systems, which utilize only information available in the CAD system, are realized. Examples show the system recognizing, locating, and tracking a variety of objects in a robot work-cell and in natural scenes.

Visual surveillance for moving vehicles

An overview is given of a vision system for locating, recognising and tracking multiple vehicles, using an image sequence taken by a single camera mounted on a moving vehicle. The camera motion is estimated by matching features on the ground plane from one image to the next. Vehicle detection and hypothesis generation are performed using template correlation and a 3D wire frame model of the vehicle is fitted to the image. Once detected and identified, vehicles are tracked using dynamic filtering. A separate batch mode filter obtains the 3D trajectories of nearby vehicles over an extended time. Results are shown for a motorway image sequence.

Vision-based vehicle detection and tracking with a mobile camera using a statistical framework

En esta tesis se aborda la detección y el seguimiento automático de vehículos mediante técnicas de visión artificial con una cámara monocular embarcada. Este problema ha suscitado un gran interés por parte de la industria automovilística y de la comunidad científica ya que supone el primer paso en aras de la ayuda a la conducción, la prevención de accidentes y, en última instancia, la conducción automática. A pesar de que se le ha dedicado mucho esfuerzo en los últimos años, de momento no se ha encontrado ninguna solución completamente satisfactoria y por lo tanto continúa siendo un tema de investigación abierto. Los principales problemas que plantean la detección y seguimiento mediante visión artificial son la gran variabilidad entre vehículos, un fondo que cambia dinámicamente debido al movimiento de la cámara, y la necesidad de operar en tiempo real. En este contexto, esta tesis propone un marco unificado para la detección y seguimiento de vehículos que afronta los problemas descritos mediante un enfoque estadístico. El marco se compone de tres grandes bloques, i.e., generación de hipótesis, verificación de hipótesis, y seguimiento de vehículos, que se llevan a cabo de manera secuencial. No obstante, se potencia el intercambio de información entre los diferentes bloques con objeto de obtener el máximo grado posible de adaptación a cambios en el entorno y de reducir el coste computacional. Para abordar la primera tarea de generación de hipótesis, se proponen dos métodos complementarios basados respectivamente en el análisis de la apariencia y la geometría de la escena. Para ello resulta especialmente interesante el uso de un dominio transformado en el que se elimina la perspectiva de la imagen original, puesto que este dominio permite una búsqueda rápida dentro de la imagen y por tanto una generación eficiente de hipótesis de localización de los vehículos. Los candidatos finales se obtienen por medio de un marco colaborativo entre el dominio original y el dominio transformado. Para la verificación de hipótesis se adopta un método de aprendizaje supervisado. Así, se evalúan algunos de los métodos de extracción de características más populares y se proponen nuevos descriptores con arreglo al conocimiento de la apariencia de los vehículos. Para evaluar la efectividad en la tarea de clasificación de estos descriptores, y dado que no existen bases de datos públicas que se adapten al problema descrito, se ha generado una nueva base de datos sobre la que se han realizado pruebas masivas. Finalmente, se presenta una metodología para la fusión de los diferentes clasificadores y se plantea una discusión sobre las combinaciones que ofrecen los mejores resultados. El núcleo del marco propuesto está constituido por un método Bayesiano de seguimiento basado en filtros de partículas. Se plantean contribuciones en los tres elementos fundamentales de estos filtros: el algoritmo de inferencia, el modelo dinámico y el modelo de observación. En concreto, se propone el uso de un método de muestreo basado en MCMC que evita el elevado coste computacional de los filtros de partículas tradicionales y por consiguiente permite que el modelado conjunto de múltiples vehículos sea computacionalmente viable. Por otra parte, el dominio transformado mencionado anteriormente permite la definición de un modelo dinámico de velocidad constante ya que se preserva el movimiento suave de los vehículos en autopistas. Por último, se propone un modelo de observación que integra diferentes características. En particular, además de la apariencia de los vehículos, el modelo tiene en cuenta también toda la información recibida de los bloques de procesamiento previos. El método propuesto se ejecuta en tiempo real en un ordenador de propósito general y da unos resultados sobresalientes en comparación con los métodos tradicionales. ABSTRACT This thesis addresses on-road vehicle detection and tracking with a monocular vision system. This problem has attracted the attention of the automotive industry and the research community as it is the first step for driver assistance and collision avoidance systems and for eventual autonomous driving. Although many effort has been devoted to address it in recent years, no satisfactory solution has yet been devised and thus it is an active research issue. The main challenges for vision-based vehicle detection and tracking are the high variability among vehicles, the dynamically changing background due to camera motion and the real-time processing requirement. In this thesis, a unified approach using statistical methods is presented for vehicle detection and tracking that tackles these issues. The approach is divided into three primary tasks, i.e., vehicle hypothesis generation, hypothesis verification, and vehicle tracking, which are performed sequentially. Nevertheless, the exchange of information between processing blocks is fostered so that the maximum degree of adaptation to changes in the environment can be achieved and the computational cost is alleviated. Two complementary strategies are proposed to address the first task, i.e., hypothesis generation, based respectively on appearance and geometry analysis. To this end, the use of a rectified domain in which the perspective is removed from the original image is especially interesting, as it allows for fast image scanning and coarse hypothesis generation. The final vehicle candidates are produced using a collaborative framework between the original and the rectified domains. A supervised classification strategy is adopted for the verification of the hypothesized vehicle locations. In particular, state-of-the-art methods for feature extraction are evaluated and new descriptors are proposed by exploiting the knowledge on vehicle appearance. Due to the lack of appropriate public databases, a new database is generated and the classification performance of the descriptors is extensively tested on it. Finally, a methodology for the fusion of the different classifiers is presented and the best combinations are discussed. The core of the proposed approach is a Bayesian tracking framework using particle filters. Contributions are made on its three key elements: the inference algorithm, the dynamic model and the observation model. In particular, the use of a Markov chain Monte Carlo method is proposed for sampling, which circumvents the exponential complexity increase of traditional particle filters thus making joint multiple vehicle tracking affordable. On the other hand, the aforementioned rectified domain allows for the definition of a constant-velocity dynamic model since it preserves the smooth motion of vehicles in highways. Finally, a multiple-cue observation model is proposed that not only accounts for vehicle appearance but also integrates the available information from the analysis in the previous blocks. The proposed approach is proven to run near real-time in a general purpose PC and to deliver outstanding results compared to traditional methods.

Multi-agent Architecture for Heterogeneous Reasoning under Uncertainty Combining MSBN and Ontologies in Distributed Network Diagnosis

This article proposes a MAS architecture for network diagnosis under uncertainty. Network diagnosis is divided into two inference processes: hypothesis generation and hypothesis confirmation. The first process is distributed among several agents based on a MSBN, while the second one is carried out by agents using semantic reasoning. A diagnosis ontology has been defined in order to combine both inference processes. To drive the deliberation process, dynamic data about the influence of observations are taken during diagnosis process. In order to achieve quick and reliable diagnoses, this influence is used to choose the best action to perform. This approach has been evaluated in a P2P video streaming scenario. Computational and time improvements are highlight as conclusions.

Publishing Orthology and Diseases Information in the Linked Open Data Cloud

The Linked Data initiative offers a straight method to publish structured data in the World Wide Web and link it to other data, resulting in a world wide network of semantically codified data known as the Linked Open Data cloud. The size of the Linked Open Data cloud, i.e. the amount of data published using Linked Data principles, is growing exponentially, including life sciences data. However, key information for biological research is still missing in the Linked Open Data cloud. For example, the relation between orthologs genes and genetic diseases is absent, even though such information can be used for hypothesis generation regarding human diseases. The OGOLOD system, an extension of the OGO Knowledge Base, publishes orthologs/diseases information using Linked Data. This gives the scientists the ability to query the structured information in connection with other Linked Data and to discover new information related to orthologs and human diseases in the cloud.

Estudos sobre a organização genômica, evolução e expressão de microRNAs

Os microRNAs (miRNAs) são pequenos RNAs não codificadores de proteínas presentes na maioria dos eucariotos. Esses RNAs regulam a expressão gênica em nível pós-transcricional através do silenciamento de mRNAs-alvo que possuem sítios complementares às suas sequências, atuando em praticamente todos os processos celulares. Embora a estrutura e função dos miRNAs estejam bem caracterizadas, aspectos relacionados à sua organização genômica, evolução e atuação em doenças são tópicos que apresentam enormes lacunas. Nesta tese, utilizamos abordagens computacionais para investigar estes temas em três trabalhos. No primeiro, processamos e integramos um vasto volume de dados publicamente disponíveis referentes aos miRNAs e genes codificadores de proteínas para cinco espécies de vertebrados. Com isso, construimos uma ferramenta web que permite a fácil inspeção da organização genômica dos miRNAs em regiões inter e intragênicas, o acesso a dados de expressão de miRNAs e de genes codificadores de proteínas (classificados em genes hospedeiros e não hospedeiros de miRNAs), além de outras informações pertinentes. Verificamos que a ferramenta tem sido amplamente utilizada pela comunidade científica e acreditamos que ela possa facilitar a geração de hipóteses associadas à regulação dos miRNAs, principalmente quando estão inseridos em genes hospedeiros. No segundo estudo, buscamos compreender como o contexto genômico e a origem evolutiva dos genes hospedeiros influenciam a expressão e evolução dos miRNAs humanos. Nossos achados mostraram que os miRNAs intragênicos surgem preferencialmente em genes antigos (origem anterior à divergência de vertebrados). Observamos que os miRNAs inseridos em genes antigos têm maior abrangência de expressão do que os inseridos em genes novos. Surpreendentemente, miRNAs jovens localizados em genes antigos são expressos em um maior número de tecidos do que os intergênicos de mesma idade, sugerindo uma vantagem adaptativa inicial que pode estar relacionada com o controle da expressão dos genes hospedeiros, e como consequência, expondo-os a contextos celulares e conjuntos de alvos diversos. Na evolução a longo prazo, vimos que genes antigos conferem maior restrição nos padrões de expressão (menor divergência de expressão) para miRNAs intragênicos, quando comparados aos intergênicos. Também mostramos possíveis associações funcionais relacionadas ao contexto genômico, tais como o enriquecimento da expressão de miRNAs intergênicos em testículo e dos intragênicos em tecidos neurais. Propomos que o contexto genômico e a idade dos genes hospedeiros são fatores-chave para a evolução e expressão dos miRNAs. Por fim, buscamos estabelecer associações entre a expressão diferencial de miRNAs e a quimioresistência em câncer colorretal utilizando linhagens celulares sensíveis e resistentes às drogas 5-Fluoruracil e Oxaliplatina. Dentre os miRNAs identificados, o miR-342 apresentou níveis elevados de expressão nas linhagens sensíveis à Oxaliplatina. Com base na análise dos alvos preditos, detectamos uma significativa associação de miR-342 com a apoptose. A superexpressão de miR-342 na linhagem resistente SW620 evidenciou alterações na expressão de genes da via apoptótica, notavelmente a diminuição da expressão do fator de crescimento PDGFB, um alvo predito possivelmente sujeito à regulação direta pelo miR-342.

Microbial Community Responses to Environmental Perturbation

Microorganisms mediate many biogeochemical processes critical to the functioning of ecosystems, which places them as an intermediate between environmental change and the resulting ecosystem response. Yet, we have an incomplete understanding of these relationships, how to predict them, and when they are influential. Understanding these dynamics will inform ecological principles developed for macroorganisms and aid expectations for microbial responses to new gradients. To address this research goal, I used two studies of environmental gradients and a literature synthesis.

With the gradient studies, I assessed microbial community composition in stream biofilms across a gradient of alkaline mine drainage. I used multivariate approaches to examine changes in the non-eukaryote microbial community composition of taxa (chapter 2) and functional genes (chapter 3). I found that stream biofilms at sites receiving alkaline mine drainage had distinct community composition and also differed in the composition of functional gene groups compared with unmined reference sites. Compositional shifts were not dominated by groups that could benefit from mining associated increases of terminal electron acceptors; two-thirds of responsive taxa and functional gene groups were negatively associated with mining. The majority of subsidies and stressors (nitrate, sulfate, conductivity) had no consistent relationships with taxa or gene abundances. However, methane metabolism genes were less abundant at mined sites and there was a strong, positive correlation between selenate reductase gene abundance and mining-associated selenium. These results highlighted the potential for indirect factors to also play an important role in explaining compositional shifts.

In the fourth chapter, I synthesized studies that use environmental perturbations to explore microbial community structure and microbial process connections. I examined nine journals (2009–13) and found that many qualifying papers (112 of 148) documented structure and process responses, but few (38 of 112 papers) reported statistically testing for a link. Of these tested links, 75% were significant. No particular approach for characterizing structure or processes was more likely to produce significant links. Process responses were detected earlier on average than responses in structure. Together, the findings suggested that few publications report statistically testing structure-process links; but when tested, links often occurred yet shared few commonalities in linked processes or structures and the techniques used for measuring them.

Although the research community has made progress, much work remains to ensure that the vast and growing wealth of microbial informatics data is translated into useful ecological information. In part, this challenge can be approached through using hypotheses to guide analyses, but also by being open to opportunities for hypothesis generation. The results from my dissertation work advise that it is important to carefully interpret shifts in community composition in relation to abiotic characteristics and recommend considering ecological, thermodynamic, and kinetic principles to understand the properties governing community responses to environmental perturbation.