PBRC Kickoff Meeting
Metabolomics—a tool to measure the metabolic phenotype of a biological system
Ute Roessner
Australian Centre for Plant Functional Genomics and Metabolomics Australia, School of Botany, The University of Melbourne
Metabolomics comprises the combination of high-throughput analytical technologies for the detection and quantification of metabolites in biological systems with the application of sophisticated bioinformatic tools for data mining and analysis. I will give an introduction to Metabolomics as a technology and will demonstrate examples where the metabolite profile has been shown to be strongly correlated with a phenotype or morphological traits in plants. In addition, a summary of current bioinformatic activities to support metabolomics through faster and more reliable software for data extraction from analytical instrumentation and for sophisticated statistical and multi-variant analysis workflows for data interpretation will be provided. The capabilities of Metabolomics Australia, an NCRIS 5.1 investment through Bioplatforms Australia Ltd, will also be summarized as well as efforts to integrate “omics” data into a Systems Biology context.
Theoretical modelling and simulation of colloidal and biological systems
Stan Miklavcic
School of Mathematics and Statistics, University of South Australia
Colloidal systems feature in a vast variety of situations from chemical industry processes (minerals processing, oil production, pulp and paper manufacturing) to commercial products (milk, soap and detergents, hair shampoos), to biological systems and even medical applications (proteins, membranes, cells).
The submicroscopic and even nanoscopic properties of colloidal systems characterise their macroscopic behaviour. Because of the enormous number of constituent molecules involved, a proper theoretical description requires a full statistical mechanics analysis; the principal ingredients in the ensuing statistical mechanical model are the intermolecular forces acting between the constituent molecules.
In this talk I will give an overview of my theoretical and experimental research on colloidal systems of general and specific relevance to biology. The overview will include mention of
-
studies of properties and behaviour of macromolecules (polyelectrolytes, proteins) in bulk and at interfaces,
-
the electrical double layer
-
surface forces (edl, van der Waals, hydration, etc.)
-
electrokinetics
Modeling the evolution of gene duplications
Andreas Schreiber
Australian Centre for Plant Functional Genomics
The duplication and deletion of existing genes in a genome is recognized as an important evolutionary mechanism for creating phenotypic diversity. It is responsible for the creation of “gene families” that are particularly common in plants. It also accounts for variation in “gene copy number”, which has quite recently been shown to be an important factor in susceptibility to disease. Furthermore, at least for bacterial organisms, mathematical modeling of the evolution of gene family sizes as a modified birth-death process has provided us with a deeper understanding of the biological data.
I will give an elementary overview of this topic, covering both biological as well mathematical aspects, and will outline some ideas on how more detailed modeling might provide benefits to plant science.
Research in Chem-, Wood- & Bio-informatics; Genomics & Environ- & Psycho-metrics
Irene Hudson and Chris Brien
School of Mathematics and Statistics, University of South Australia
Five current research areas in informatics and metrics: theory and application are briefly discussed.
-
Cheminformatics: Bayesian Multivariate Mixture (BMM) models: ARC DP (2007-2010) High throughput docking of small molecule ligands into high resolution protein structures is standard in computational approaches to drug discovery. We are creating new indicators of molecular ligand binding for drug discovery, with a focus on developing novel Bayesian (theory and methods for) classification. This is being tested on calpain inhibitors for cataract treatment. Our new indicators provide alternatives to diagnostics currently used in molecular libraries (Kim, Hudson et al, 2008). Open problem: To test the predictive accuracy and robustness of our Bayesian methods by accessing alternative molecular libraries. ARC research with Adelaide University and GKSS, Berlin.
-
Genomics: Statistical & computational methods for genetic association with diseases: 2009- Whole genome data promises to provide increased detail about genetic variation in relation to diseases. We shall develop new methods to ID disease specific gene markers via reductive methods for high-dimensional (SNP arrays). Open problems: What are the most optimal models which mirror traditional point-wise statistics developed to date? Can likely candidate genes be identified from biological knowledge? What design / proximity issues pertain in gene ID? Can we develop our Bayesian methods here? Will variants of our mixture models (Hudson, Kim et al.) work on the whole genome data? This research is in collaboration with CSIRO P-Health and Oxford University.
-
Paleo-climatic interpretation at a within tree ring level - mapping of multiple climate proxies: Are within ring wood anatomy and flowering records proxies for global climate change? The wealth of information on climate signals embedded within tree ring structure remains untapped and awaits rigorous investigation. We are utilising a UniSA ARC DP development grant to obtain high resolution wood anatomical, climate, flowering & tree stem growth data from Eucalypt trees in Vic. Data will then be mapped onto high throughput SilviScan wood fibre profiles obtained from the same trees by CSIRO. Open problems: Climate cannot be reconstructed from intra-ring wood profiles or dendrometer growth band data, unless we can rigorously map the cell and tree growth distance scales (d2) to the chronological time scale (t1) of climate and vice versa. We shall develop an optimization pathway, a so-called time-space analysis of the recurrent states in multivariate non-stationary time series. Sampling & design? Collaboration with CSIRO Forestry, Melbourne University & Vienna.
-
Phenological statistical techniques and climate change: evidence based methods: The impacts of climate change are documented in the Northern Hemisphere, where phenology (flowering time, birds) provide much of the basis of reported climate impacts (2001 & 2007 IPCC). There is however, a paucity of phenological recording in Australia (Hudson et al., 2008; Keatley & Hudson, 2005). No Australian contribution appeared in the 2001 IPCC report. Our work on phenological statistics, with application to a unique Southern Hemisphere Eucalypts flowering record, is developing Southern Hemisphere phenological climate proxies, an acknowledged national priority. This research aims to create rigorous statistical methods for (evidence-based) climate change, and will be a major part of a forthcoming Springer book “Phenological Research: Mathematical and Statistical Methods, Design & Applications (2009).” Open problem: Are herbarium collections / photos able to accurately track phenological shifts in relation to climate? UniMelb, UniAdelaide, SA State Herbarium collaboration.
-
Psychometrics: ICA on brain image data and gender specific prediction of depression: A major question in psychometrics is whether personality affects mental illness. The psychobiological model for temperament and character (TCI) of Cloninger (2002) offers an approach to this question. Our ICA maps (Kang & Hudson, 2009) of brain function onto TCIs reveal that personality is correlated with specific brain regions. This mathematical evidence supports a possible biological basis for depression. Earlier Turner, Hudson et al. (2003) implemented a quartile adaptation at a voxel-by-voxel level, which allowed testing of non-linearity of blood flow across TCIs. Open problem: Can we derive a GAMLSS application to voxel based modelling of brain image data (Hudson & Brien et al., 2009)? What stats design to use? Collaboration with the University of Canterbury (UC) and Sydney University.
Innovative methods for time series modelling of climate variables
John Boland
School of Mathematics and Statistics, University of South Australia
We model the level and volatility of climate variables, specifically solar radiation and wind speed using non-standard time series techniques. They are non-standard only in that they are normally not used for modelling these variables. In fact, it is not wind speed per se that we model, but rather the output from single wind farms or a collection of wind farms.
In this presentation, I will focus on forecasting the level and volatility of wind farm output series. There are various approaches to forecasting climate variables, using alternatively, Markov models, state space models, neural nets or Box-Jenkins methods. We use the latter, coupled with models of the deterministic component – identified using spectral analysis - which we also denote as classical time series modelling structures. There is a specific reason for choosing this methodology. It is the approach that gives the most knowledge of the underlying physical nature of the phenomenon. We identify the various components inherent in the time series, seasonality, autoregressive structure, and the statistical properties of the noise.
The traditional time series approach described above is of great value, but may miss some of the inter-dependence of the volatility terms. Financial time series analysis has focused on specific models for this volatility, using techniques described as auto-regressive conditional heteroscedastic (ARCH) and Generalised ARCH (or GARCH) models and their variants. Essentially, these are used to describe series that have weak or no auto-regressive structure but still display dependence. These models are normally applied to series such as stock market indices, but we use them for wind farm output time series. I will describe how the methods adopted from financial time series can aid in forecasting the volatility for wind farm output.
Between greenhouse and computer—small scale phenomics using the LemnaTec Scanalyzer
Bettina Berger and Karthika Rajendran
Australian Centre for Plant Functional Genomics
The LemnaTec Scanalyzer takes non-destructive 3D images, which are used to monitor the phenotype of barley and wheat grown in the greenhouse. Side and top view images are taken in regular intervals to follow the growth and development of the plants. These images are then analyzed using the LemnaTec software to extract biological meaningful data such as projected leaf area or degree of leaf damage, which allows conclusions about the plants performance under the chosen growth conditions. Whereas the data acquisition and image analysis are making good progress, the fraction of data used for phenotypic characterization of the plants is still limited to few parameters.
We will outline the workflow and analysis of a phenotyping experiment and discuss the problems encountered with data storage, handling and processing.
Anomaly detection in high-dimensional data sets
Belinda Chiera
School of Mathematics and Statistics, University of South Australia
A major stumbling block in real-time anomaly detection is that the volume of available data is typically high dimensional, noisy, incomplete and often non-time homogeneous, making efficient and meaningful detection onerous at best. This talk presents an overview of challenges and recent advances in this area including: feature selection; dimensionality reduction; and a "real-time" anomaly detection scheme stemming from a judicious combination of a subspace- based model-free state-space Kalman filter, a cepstral information norm and distance metrics defined over the data. An example taken from recent work will demonstrate the highlights and pitfalls of the anomaly detection scheme over a top level Internet topology in the midst of an anomalous event, followed with a discussion of open issues in this area.
ACPFG proteomics capability: analytical & bioinformatic
Andrew Cassin
Australian Centre for Plant Functional
Genomics
Proteomics is the analysis of the protein complement of a biological system.
For many years it has suffered from the lack of a quantitative approach that
allows comparison of data from different samples within a biological
experiment. Recently, the development of iTRAQ and MRM techniques has
allowed this type of approach. I will present an introduction to iTRAQ
protein quantitation via the current investigation of a proteomics study of
the wheat drought experiments being conducted in Adelaide and for which both
transcriptomics and metabolomics data are being gathered. Demonstration of a
Proteomics Drought Comparator bioinformatics website including preliminary
results, current issues and future work will be presented.
Design and Analysis of Experiments
Chris Brien
School of Mathematics and Statistics,
University of South Australia
My research area is the design and analysis of experiments. I will outline
my recent work on 1) designs for multifactor, glasshouse experiments to
assess indigenous plant species, 2) the characterization and design of
experiments with multiple randomizations, and 3) the mixed model analysis of
experiments, including multitiered and longitudinal experiments. I will also
outline proposed future research.
Making fuel- and drug-producing microbes through analysis, modelling, and design
Desmond Lun
School of Mathematics and Statistics,
University of South Australia and Australian Centre for Plant Functional
Genomics
Engineering microorganisms that efficiently produce drugs and fuels is an
exciting and challenging problem with large potential impact on energy
supply, the environment, and global health. Such engineering is greatly
aided by systematic design and, in this talk, we discuss how systematic
design can be achieved through the analysis and modelling of microbial
metabolic networks. We discuss approaches that we are developing for
modelling metabolism and gene regulation and for using these models to guide
design. In particular, we describe a network optimization problem that
arises in the context of optimal design and discuss algorithmic approaches
for its solution. We describe our progress in engineering E. coli for
petroleum production from simple sugars and in other engineering directions.