The series aims to foster knowledge exchange at the intersection of biology, statistics, computing, and data science. We hope to use this as a regular platform for the community to share their work, spark meaningful discussions and cultivate interdisciplinary collaborations.
To facilitate participation from across campus, the seminars will primarily be held online via Zoom. If you are interested in organising or joining a watch party, please get in touch.
- Where
- Online via Zoom
- Watch parties
- Get in touch to organise or join one
- Organiser
- Dr Weihao (Patrick) Li
Talks
7 Sept 2026 11am
Investigating genotype by environment interaction effect in the genomic prediction of rain-grown cotton
Justin Borevitz, Australian National University
Abstract and speaker bio
Prof. Justin Borevitz will present his team's latest work on genetically diverse lines of wheat, barley, canola and even eucalyptus. They use long read sequencing, genome assembly and SNP/SV genotyping to develop and select powerful mapping populations. These are grown in contrasting simulated and field seasonal conditions for developmental phenotyping using cameras, drones and satellites. Genome wide association studies map the adaptive genetic basis for complex traits while controlling for neutral genome wide relatedness. Finally genomic prediction is used to select ideal lines for novel environments. He will present case studies to explain the methods.
About the speaker Justin Borevitz is a Professor in the Research School of Biology at the Australian National University, where he leads the Borevitz Group, which studies plant genomics and climate adaptation. He received his PhD from the University of California, San Diego, in 2002 and subsequently conducted postdoctoral research at the Salk Institute. He was an Assistant and Associate Professor in the Department of Ecology and Evolution at the University of Chicago before joining ANUe. His research combines population genomics, genome-wide association studies, landscape genomics, and high-throughput phenotyping to understand the genetic basis of plant adaptation to changing environments. His work spans model plants, crops, and foundation species such as Eucalyptus, with the goal of identifying adaptive genetic variation and using genomic and phenomic approaches to support climate-resilient agriculture, conservation, and ecological restoration.
27 July 2026 11am
Investigating genotype by environment interaction effect in the genomic prediction of rain-grown cotton
Qian Feng, CSIRO agriculture and food unit
Abstract and speaker bio
Quantitative trait variance is attributed to genotype (G), environment (E) and the interaction of genotype and environment (G × E) effects. The existence of G × E effects indicate the individuals with the same genotype respond differently under different environment conditions. Identifying such effects is important from the agronomic management perspective, but also from a plant breeding perspective where the objective is to identify genetically elite lines. In this study, we specifically focus on rain-grown cotton and investigate the G × E effects for multiple fibre yield and fibre quality traits. We use a large-scale dataset collected over 23 years at 4 growing sites in central eastern Australia that various traits, genotype and climate data are available. We first use eight complementary statistical methods (including multi-layer group-lasso and random forest) to determine the influential environmental covariates associated with each trait. We next model the trait performance using genomic prediction models with and without G × E effects. Our results suggest that environmental covariates have a significant impact on the performance of genomic prediction models, where some environmental covariates simultaneously affect multiple traits. Conversely, limited G × E effects are identified for most of the traits analysed. In this talk, the results of this study will be interpreted to determine the potential use of genomic prediction tools in breeding decisions when developing new cotton varieties.
About the speaker Qian Feng is a CERC research fellow at CSIRO agriculture and food unit. Her current research focuses on designing and implementing statistical methods to accelerate the breeding of Australian rain-grown cotton.
22 June 2026 11am
Hidden Mappings in the Grammar of Graphics
Paul Murrell, University of Auckland
Abstract and speaker bio
The Grammar of Graphics provides a framework for efficiently specifying a data visualisation. A plot can be described in terms of aesthetic mappings - transformations of data values to the visual attributes of geometric shapes. For example, a bar plot is a mapping of counts to the lengths of bars. In this talk, we will explore both a simplification of and an extension of the Grammar of Graphics, in both cases by recasting disparate features of a data visualisation as just further examples of aesthetic mappings.
About the speaker Paul Murrell is an Associate Professor in the Department of Statistics at the University of Auckland. His research focuses on software for producing statistical graphics and he is a member of the R core development team for the R Project. He also teaches courses on Statistical Computing and Data Visualisation.
18 May 2026 11am
DIVAS: finding shared and individual variation in multimodal data
Jiadong Mao, University of Melbourne
Abstract and speaker bio
Many modern studies generate several data matrices measured on the same samples, e.g. different molecular assays, different imaging modalities, different sensors, different views of the same units. A natural question is which signals are shared across all data blocks, which are confined to specific subsets, and which are unique to individual blocks. Most popular integration methods capture mainly jointly shared or individual variation, leaving partially shared structure underexplored. DIVAS (Data Integration Via Analysis of Subspaces) addresses this gap by systematically decomposing matched data blocks into jointly shared, partially shared and modality-specific components with formal statistical inference. In this talk I will introduce our open-source R package implementing DIVAS with improved usability and visualisation tools, then walk through a bioinformatics case study. We integrate six blocks of COVID-19 patient data including single-cell RNA-seq from four immune cell types, proteomics and metabolomics. DIVAS recovers partially shared components that track disease severity and reveal coordinated immune–metabolic dysregulation invisible to joint-only methods. The tool is general and applies wherever data can be organised into matched matrices. Joint work with Yinuo Sun, Kim-Anh Lê Cao and J. S. Marron.
About the speaker Dr Jiadong Mao is a statistician and postdoctoral researcher at Melbourne Integrative Genomics, University of Melbourne. He develops statistical and machine learning methods for complex biomedical ‘omics’ data, with a focus on modelling cell states and uncovering molecular drivers of disease progression. His work spans single-cell, spatial and longitudinal omics data. More broadly, he works at the interface of statistical methodology, cell biology, multiomic integration and clinical translation, with collaborations spanning cancer, neuroscience and longitudinal clinical studies.
30 Mar 2026 11am
ggincerta: Extend 'ggplot2' with Layers and Scales for Spatial Uncertainty Visualisation
Xueqi (Maggie) Ma, Australian National University
Abstract and speaker bio
Uncertainty is inherent in spatial data, yet it is often overlooked in visualisation, leading to decisions based solely on point estimates. Existing tools for visualising spatial uncertainty, such as the Vizumap R package, provide several approaches, including bivariate choropleth maps, pixel maps, glyph maps, and exceedance probability maps, but are not implemented as standard extensions of ggplot2. We introduce the ggincerta R package, a ground-up reimplementation that recreates Vizumap’s visualisations within the layered grammar of graphics framework. By leveraging ggplot2, ggincerta integrates uncertainty visualisation into existing workflows and works seamlessly with other ggplot2 components using familiar syntax and conventions.
About the speaker Xueqi (Maggie) Ma is a PhD student in Statistics at the Australian National University. Her research focuses on the visualisation of spatial uncertainty within the ggplot2 framework. She is developing the R package ggincerta, which provides specialised ggplot2 layers and scales for visualising uncertainty in areal data.
17 Nov 2025 11am
Predicting changes in agricultural yields under climate change using meta-analytic mixed effects models
Christine Li & Tom Kompas, University of Melbourne
Abstract and speaker bio
Mixed effects models trained on data collected from meta-analysis can be used to generate expectations of crop yield responses to climatic factors. In this talk, we present methods and results from our recent study that updates published estimates of crop yield responses to projected temperature, precipitation, and CO2 patterns. Our objectives were to model potential climate impacts in agriculture using a statistical modelling approach, for input into large-dimensional climate and trade modelling, and to systematically evaluate modelled uncertainty across multiple dimensions (e.g. model choice, sampling, climate projection data and emissions scenarios) using a block-bootstrapping approach. We show that mixed effects models perform better than pooled OLS models on root mean squared error (RMSE) and explained deviance, despite the common usage of pooled OLS in previous meta-analyses. Based on our analysis, the use of pooled OLS may underestimate yield losses. Our estimates show projected yield responses of − 22% (maize), − 9% (rice), − 15% (soy), and − 14% (wheat) from 2015 to 2080–2100 under the business-as-usual scenario of SSP5–8.5, which reduce to − 3.8%, − 2.7%, 1.4%, and − 1.5% respectively under the lower emissions scenario of SSP1–2.6.
About the speaker Christine Li is a PhD Candidate at the Centre of Excellence for Biosecurity Risk Analysis with a background in agricultural, ecological and biosecurity economics. Her research interests include spatial mapping and valuation of ecosystem goods and services, statistical meta-analysis and modelling of agricultural losses and damages from climate change, mitigation and adaptation measures. Her applied work in the biosecurity sector includes valuing assets at risk from pests and diseases, cost-benefit analysis and supporting the development of decision support tools. Tom Kompas is a Professor of Environmental Economics and Biosecurity in the School of Biosciences and the School of Ecosystem and Forest Sciences at the University of Melbourne. He is also a Chief Investigator in the Centre of Excellence for Biosecurity Risk Analysis (CEBRA) and Research Group Director of the Centre for Environmental and Economic Research (CEER) at the University of Melbourne. Tom is a Fellow of the Academy of Social Sciences Australia, a recipient of the Eureka Prize in Science and a Distinguished Fellow of the Australasian Agricultural and Resource Economics Society. He was Director of the Crawford School of Public Policy at the Australian National University from 2009 to 2015.
20 Oct 2025 11am
A predictive model for genotype x environment x management practice (GxExM) interactions incorporating environmental covariates
Michael Mumford, Queensland Department of Primary Industries
Abstract and speaker bio
In field crops research, genotype by environment (G×E) interactions are ubiquitous. The development of statistical methods to model the genotype by environment interaction using environmental covariates (ECs) has mostly occurred within the context of crop breeding. In agronomic research, measuring the impact of management practice (M) on genotype performance is also a key objective, giving rise to the genotype by environment by management practice (G×E×M) interaction. A one-stage analysis approach, implemented in a linear mixed model framework, is presented, which incorporates ECs to disentangle the G×E×M interaction effects. The linear mixed model framework allows adjustments for design effects and spatial field trend, along with the estimation of heterogeneous residual variance across environments. An application of the methodology will be demonstrated using a series of sorghum agronomy field experiments. Recently, a complementary R-package called ecreml was developed to enable researchers to easily apply the developed statistical methodology to their own multi-environment trial data consisting of G×E×M interaction effects, and is publicly available to install via GitHub.
About the speaker Michael is a biometrician at the Queensland Department of Primary Industries, with extensive experience collaborating with agricultural researchers on a range of Grain Research Development Corporation (GRDC) projects; initially through the Statistics for the Australian Grains Industry (SAGI) project and more recently the Analytics for the Australian Grain Industry (AAGI) initiative. Michael’s research interests are in statistical methodological development, with a focus on developing methodologies that are directly applicable to agricultural experiments. Michael is also completing PhD studies part-time currently through the University of Queensland, focused on developing improved statistical methodologies for the analysis of count data arising from comparative agricultural experiments using generalised linear mixed models.
7 Oct 2025 3pm
Design and Spatial Analysis of On-Farm Strip Trials: From Principles to Simulation Evaluations
Zhanglong Cao, Curtin University
Abstract
This study explores statistical strategies for the design and analysis of on-farm experiments (OFE), grounded in established principles of experimental design and spatial modelling. Through simulation studies, we assess various approaches, including design layouts combined with Geographically Weighted Regression (GWR) for continuous response variables and Linear Mixed Models (LMM) for categorical treatments. Our results align with key findings in the literature, emphasising the importance of appropriate design and modelling choices for effectively capturing spatial heterogeneity. Additionally, we compare two trial types—large-strip trials and stacked replicated trials—and highlight the significance of data granularity, which informs data collection strategies for industry partners. The findings advocate for appropriate trial designs for different purpose, and increased within-trial sampling, rather than reliance solely on average values to better reflect local variability and enhance inference accuracy. This work underscores the critical role of tailored statistical methods in improving the reliability and practical applicability of OFE outcomes.
15 Aug 2025 2pm
How will future climate change affect wheat yield and nutritional quality?
Zixiong Zhuang, Australian National University
Abstract and speaker bio
In the coming decades, climate change will continue to bring higher atmospheric CO2 concentrations as well as warmer air temperatures. However, the combined effects of elevated CO₂ (eCO2) and elevated temperature (eT) on wheat grain yield and nutrition quality remain largely unknown. When tested separately, eCO2 increases wheat yield and decreases grain nutritional quality, whereas eT decreases yield but with unknown effects on grain quality. To better understand the concurrent treatment of eCO2 and eT on wheat performances as per realistic future climate conditions, this talk presents early insights into one of the largest experiments to expand our knowledge in this area — testing 90 genetically diverse wheat lines under projected climate conditions by the year 2100 (800 ppm CO2 and + 5°C). This talk presents preliminary results from the first year’s experiment of this project. I outline both the traditional biometrics and high-throughput phenotyping (HTP) traits captured to explain the variations in plant responses to future climate conditions. I demonstrate analyses that support early understanding about the underlying biological mechanisms contributing to variations in wheat yield and nutrition qualities that can be studied further in future experiments.
About the speaker Zixiong is a second-year PhD student at the Research School of Biology (RSB). He is supervised by Professor Danielle Way where he focuses on plant physiological responses to future climate change in Australian spring wheat lines. Zixiong completed his Master’s degree with Professor Justin Borevitz at the Australian National University. During Zixiong’s Master’s degrees, he discovered genomic structural variants using population-level long-read sequencing on wild Eucalyptus pangenomes. He joined the Way Lab in 2023, transitioning his focus from population genomics to ecophysiology and plant phenomics.
1 Aug 2025 2pm
Novel methods for analysing single cell gene expression data
Yidi Deng, Australian National University
Abstract and speaker bio
This talk will introduce three novel methods we developed to address key challenges in single-cell RNA-seq (scRNA-seq) data analysis. Sincast is an imputation method that denoises scRNA-seq data to enhance the robustness of downstream statistical analyses. Its utility is highlighted through comprehensive benchmarking of cell identity against bulk transcriptomic references. StableMate is a stability-focused regression framework that identifies functional dependencies in gene expression that are either conserved or meaningfully variable across biological contexts. When applied to scRNA-seq data, StableMate enhances reproducibility and provides regulatory insights into co-expression analyses. NeighbourNet constructs cell-specific gene regulatory networks, facilitating the fine-grained characterization of gene regulatory dynamics at the single-cell level and offering a new perspective on analysing scRNA-seq through cellular variation in co-expression.
About the speaker Yidi Deng is a statistical bioinformatician and new post-doctoral fellow, working with Dr Emi Tanaka to uncover genomic insights that support Australia’s grain-industry research. Yidi completed a PhD in applied statistics at the University of Melbourne, where he developed novel methods for multi-omics data integration and variable selection, aimed at deciphering gene-regulatory mechanisms in the context of developmental biology and immunology. Yidi’s overall research interests is to translate advanced statistical approaches into intuitive, practical software that biologists can easily adopt without specialized statistical or computational expertise.