ePosters

Image-based Predictive Modelling for the Characterisation of Cellular Senescence

Senescence is a cellular stress response that is hypothesised to act as the causal nexus of ageing through its extensive pro-inflammatory secretory phenotype. Cells entering senescence undergo extensive morphological, transcriptional and metabolic changes, which vary significantly by cell-type, inducing stressor, and time since induction. Due to its dynamic and heterogeneous nature, a specific and universal biomarker of senescence remains unidentified, presenting a major obstacle in the research and development of therapeutics.
Here we present the ongoing development of an image-based predictive model for senescence using deep convolutional neural networks and high-content microscopy of ageing MSCs, and their subsequently derived osteocytes, adipocytes and chondrocytes. Images are corrected, segmented and labelled in a pre-processing pipeline developed in Python 3, to be fed as training data to a supervised learning algorithm. Through post-hoc interpretability and visualisation methods, the final model can be analysed to provide meaningful insight into the complex relationships and features that phenotypically characterise the sub-types and transitional states of senescence. This will facilitate the identification of novel, robust biomarkers for improved therapeutic targeting. The models predictive classification of senescence from brightfield images alone will also make it a practical tool for assessing population quality in research and clinical settings.

Author: Ms Ebony Watson, The University of Queensland
Co Author/s: Dr Atefeh Taherian Fard, Dr Farhad Soheilmoghaddam,Professor Ernst Wolvetang, The University of Queensland
Associate Professor Jessica Mar, University of Southern Queensland

Integrative analysis of long non-coding RNA/pseudogene-mediated ceRNA crosstalk in endometrial cancer

Endometrial cancer (EC) is the most common gynaecological cancer among Australian women. The regulatory mechanisms of non-coding RNAs in EC are still elusive. The competing endogenous RNA (ceRNA) hypothesis states that RNAs transcripts as long non-coding RNAs (lncRNAs) and pseudogenes, can become microRNAs sponges regulating gene expression [1]. We built a ceRNA network to explore the regulatory mechanism of ceRNAs in EC. As input, we selected 3584 and 339 differentially expressed mRNAs and non-coding RNAs, respectively using TCGA-UCEC. Analysis was conducted using GDCRNATools R package, and results suggest that 7 mRNAs and 11 miRNAs are influenced by both lncRNA-miRNA-mRNA and pseudogene-miRNA-mRNA networks. Functional enrichment analysis identified that ceRNA-associated mRNAs are involved in EC-related signalling pathways like PI3K-Akt and MAPK. Univariate Cox (proportional-hazards) regression revealed that 13 lncRNAs, 3 pseudogenes, 17 microRNAs and 37 mRNAs were significantly correlated with overall survival of EC patients (p<0.01). To the best of our knowledge, this is the first comprehensive analysis of pseudogenes in EC-associated ceRNA networks. This ceRNA network provides candidate molecular biomarkers for evaluating the prognosis of EC, that will contribute to the understanding of ceRNA mechanisms involved in EC.
1. Salmena, L., et al. 2011. Cell.
2. Li, R., et al. 2020.Bioinformatics.

Author: Ms Dulari Jayarathna, Queensland University of Technology
Co Author/s: Dr Miguel Renteria, QIMR Berghofer Medical Research Institute
Associate Professor Emilie Sauret, Associate Professor Jyotsna Batra, Dr Neha Gandhi, Queensland University of Technology

Adaptive introgression revealed in Cryptosporidium hominis in Africa by comparative genomics

Cryptosporidiosis is a major cause of diarrhoeal illness among children. Cryptosporidium hominis is the dominant pathogen in Africa, and genotyping at glycoprotein-60 (gp60) gene has revealed a complex distribution of different subtypes. However, a comprehensive exploration of the metapopulation structure and evolution based on whole genome data has yet to be performed. Here, we sequenced and analysed 26 C. hominis genomes, representing different gp60 subtypes, collected at rural sites in Gabon, Ghana, Madagascar and Tanzania.
Phylogenetic and cluster analyses showed that isolates predominantly clustered by their country of origin, irrespective of their gp60 subtype. We found a significant isolation-by-distance signature that shows the importance of local transmission, but we also detected evidence of genetic introgression between isolates of different geographic regions. We identified 37 outlier genes with exceptionally high nucleotide diversity, found more often than expected in recombinant regions, and they show a distinct signature of adaptive selection.
In summary: 1) metapopulation structure of C. hominis can only be accurately captured by whole genome analyses; 2) local anthroponotic transmission underpins the spread of this pathogen in Africa; and 3) genetic introgression between distinct geographical lineages provides novel substrate for positive- or balancing selection in genes involved in host-parasite coevolution.

Author: Mr Swapnil Tichkule, The University of Melbourne

A statistical approach for modelling differential distributions in single-cell transcriptomic data

“We present a novel statistical framework for identifying differential distributions in single-cell RNA-sequencing (scRNA-seq) data between treatment conditions by modelling gene expression read counts using generalized linear models. We model each gene independently under each treatment condition using the error distributions Poisson, Negative Binomial, Zero-inflated Poisson and Zero-inflated Negative Binomial with log link function and model-based normalization for differences in sequencing depth. Model selection is done by calculating the Bayesian Information Criterion and likelihood ratio test statistic.
While most methods for differential gene expression analysis aim to detect a shift in the mean of expressed values, single-cell data are driven by over-dispersion and dropouts requiring statistical distributions that can handle the excess zeros. By modelling gene expression distributions, our framework can identify subtle variations that do not involve the change in mean. It also has the flexibility to adjust for covariates and perform multiple comparisons while explicitly modelling the variability between samples.
Through simulation, we show that this framework is able to detect zero-inflated genes and when applied to a scRNA-seq dataset on ageing, our framework was able to identify genes and pathways linked to ageing that were not discovered through traditional analysis of transcriptomic data.

Author: Miss Malindrie Dharmaratne,The University of Queensland
Co Author/s: Dr Atefeh Atefeh Taherian Fard,Associate Professor Jessica Mar, The University of Queensland
Dr Ameya Kulkarni, Albert Einstein College of Medicine, Bronx, NY

Deconvolving clones in multiclonal malaria infections using Oxford Nanopore long read sequencing

“Malaria is a potentially deadly tropical disease that, despite intensive elimination and control efforts, led to an estimated 228 million cases and 405,000 deaths in 2018 alone. Malaria is caused by infection with parasites of the Plasmodium genus. Multiclonal infections, which involve the co-circulation of multiple genetically distinct parasite clones, are common, particularly in high transmission settings. Deconvolving parasite clones from whole genome sequencing data remains an analytical challenge, with implications for the estimation of drug and treatment efficacy. By spanning multilocus haplotypes, long read sequencing can streamline the reconstruction of clonal haplotypes.

Here, we present a simple heuristic clustering method to deconvolve clonal haplotypes using Oxford Nanopore long-read sequencing, focusing on highly heterozygous regions of the parasite genome. Treating each read as a separate haplotype, we adopt a read editing approach to remove sequencing artefacts and focus on informative variation. We then construct a series of networks to assign haplotypes into clonal clusters and examine variation both within and between clusters. We apply our method to a case of treatment failure in a returning traveller to confirm a recrudescence, resulting from the expansion of the original parasite population within the patient at baseline.”

Author: Miss Somya Mehra, Burnet Institute
Co Author/s: Dr Zahra Razook, Associate Professor Alyssa Barry, Deakin University

TDAview: an online visualization tool for topological data analysis

TDAview is an online tool for topological data analysis (TDA) and visualization. It implements the Mapper algorithm for TDA and provides extensive graph visualization options. TDAview is a user-friendly tool that allows biologists and clinicians without programming knowledge to harness the power of TDA. TDAview supports an analysis and visualization mode in which a Mapper graph is constructed based on user-specified parameters, followed by graph visualization. It can also be used in a visualization only mode in which TDAview is used for visualizing the data properties of a Mapper graph generated using other open-source software. The graph visualization options allow data exploration by graphical display of metadata variable values for nodes and edges, as well as the generation of publishable figures. TDAview can handle large datasets, with tens of thousands of data points, and thus has a wide range of applications for high-dimensional data, including the construction of topology-based gene co-expression networks.

Author: Mr Kieran Walsh, The University of New South Wales
Co Author/s: Dr Mircea Voineagu, Dr Fatemeh Vafaee, Associate Professor Irina Voineagu, The University of New South Wales

Fidelity of translation initiation is required for coordinated respiratory complex assembly

Mammalian mitochondrial ribosomes are unique molecular machines that translate 11 leaderless mRNAs; however, it is not clear how mitoribosomes initiate translation, since mitochondrial mRNAs lack untranslated regions. Mitochondrial translation initiation requires the formation of a ternary complex of fMet-tRNAMet, mRNA and the 28S subunit in addition to the binding of two initiation factors: MTIF2, which closes the decoding center and stabilizes the binding of the fMet-tRNAMet to the leaderless mRNAs, and MTIF3, whose role is not clear. We show that MTIF3 is essential for survival and that in mice lacking MTIF3 there is increased but uncoordinated mitochondrial protein synthesis, resulting in loss of specific respiratory complexes. RNA-Seq analysis revealed that the 5′ regions of mitochondrial transcripts were increased, and that the stability of mt-mRNAs progressively decreased in a 5′-3′ orientation such that the read coverage at 5′ mRNA ends was increased and the 3′ ends were decreased in the knockout mice, suggesting that MTIF3 might be required for efficient translation, and therefore protection, of mt-mRNAs. Proteomic and respiratory complex assembly analyses identify that MTIF3 is required for coordinated assembly of OXPHOS complexes in vivo.

Author: Miss Danielle Rudler, The Harry Perkins Institute of Medical Research
Co Author/s: Miss Laetitia Hughes, Dr Stefan Siira, Professor Oliver Rackham, Professor Aleksandra Filipovska, Dr Kara Perks, Dr Tara Richman, Miss Irina Kuznetsova, Miss Judith Ermer, Miss Anne-marie Shearwood, The Harry Perkins Institute of Medical Research
Dr Laila Abudulal, Dr Helena Viola, Professor Livia Hool, The University of Western Australia

Cross-domain network inference reveals host-microbial crosstalk in the respiratory tract of healthy newborns

Following birth all mucosal surfaces, including our lungs, get colonised by a variety of microbes which we refer to as our microbiome. There is increasing evidence that the airway microbiome plays a key role in the establishment of respiratory health by interacting with the developing immune system early in life. To-date, studies of the field primarily focused on bacteria. But rather than being isolated, microbes are part of complex ecological niches including multiple life kingdoms (bacteria, fungi, viruses). To harness the impact of early-life trans-kingdom associations on the developing respiratory immune system, we analysed both bacterial (16S) and fungal (ITS) amplicon sequencing data generated from nasal and oral swab samples from 121 healthy newborns in conjunction with host nasal transcriptomics data from the same patients. Using SParse InversE Covariance Estimation for Ecological Association Inference (SPIEC-EASI) on multi-kingdom data, we highlighted differences in ecological network organization between nasal and oral respiratory niches. Further extension of the network to the nasal transcriptome allowed the detection of discrete host-microbial interactome clusters with distinct immunological functions. Cross-domain association networks integrating multiple life kingdoms could be the first step towards understanding how early host-microbiome interactions in the airways can impact future respiratory health.

Author: Ms Celine Pattaroni, Monash University

Long-time dynamics of a Diffusive Epidemic Model with Free Boundaries

We determine the long-time dynamical behaviour of a reaction-diffusion system with free boundaries, which models the spreading of an epidemic whose moving front is represented by the free boundaries. The system reduces to the epidemic model of Capasso and Maddalena CapassoM1981 when the boundary is fixed, and to the model of Ahn et al.Ahn2016 if diffusion of the infective host population is ignored. We prove a spreading-vanishing dichotomy and determine exactly when each of the alternatives occurs. If the reproduction number $R_0$ obtained from the corresponding ODE model is no larger than 1, the epidemic modelled here will vanish, while if $R_0>1$, the epidemic may vanish or spread depending on its initial size, determined by the dichotomy criteria. Moreover, when spreading happens, we show that the expanding front of the epidemic has an asymptotic spreading speed, which is determined by an associated semi-wave problem.
Ahn2016 I. Ahn, S. Baek, Z. Lin, The spreading fronts of an infective environment in a man-environment-man epidemic model, Appl. Math. Model., 40(2016), 7082-7101.CapassoM1981 V. Capasso, L. Maddalena, Convergence to equilibrium states for a reaction-diffusion system modelling the spatial spread of a class of bacterial and viral diseases, J. Math. Biol., 13 (1981/82), no. 2, 173-184.

Author: Miss Rong Wang, The University of New England
Co Author/s: Professor Yihong Du, The University of New England

Tumour Protocognition

Cancer is a disease caused by abnormal cell growth in a part of the body. This uncontrolled cell growth leads to lose of cell equilibrium in a tissue. Due to limited natural resources, most cells have been starved in this crowded domain. Hence cells try to escape from this stressful environment to obtain sufficient natural resources. Activities of tumour cells such as surviving in the tumour, resistance for treatments, cell plasticity, invasion and metastasis proves how tumour cells take decisions according to the environment where they live. Hence the behavior of tumour cells exactly shows proto-cognitive abilities. Understating the interaction between tumour cells and the tumour microenvironment (TME) is a clue to block or slow down the spread of cancer cells and to develop better treatment strategies. In this study, a spatial stochastic model is developed to study proto-cognitive abilities of the tumour cells. To conquer from the battle of cancer, it is important to identify strengths and weaknesses of the tumour cells. This model will provide a bridge to study intra-tumoural communication through individual tumour cell behavior under cellular stress.

Author: Mrs Hasitha N.Weerasinghe,Queensland University of Technology
Co Author/s: Professor Andrew Adamatzky,University of the West of England, Dr Petra Gener,VHIR Vall d Hebron Research Institute
Professor Sunil Lakhani, The University of Queensland, Dr Cao Son Tran, Ms Sophie Taylor, Mr Alexander Hasson, Dr Pamela M. Burrage, Professor Kevin Burrage, Associate Professor Dan V. Nicolau Jr.Queensland University of Technology.

Investigating the Role of microRNAs in Driving Glioblastoma Cell States

Glioblastoma is a highly aggressive and incurable brain cancer that affects patients of all ages. This cancer exhibits a high degree of inter and intra-patient heterogeneity, present on both genetic and transcriptomic levels, and is considered one of the major challenges impeding the development of more effective treatments. One of the factors underlying this heterogeneity is cellular plasticity, where various factors can lead cells to occupy distinct molecular states with very different behaviours. The consequence of this is a complex tumor environment comprised of subpopulations of cells with different growth rates, metastatic potential or vulnerability to certain treatments. Our work is investigating how miRNAs – small non-coding RNAs which regulate post-transcriptional regulation of nearly all genes – may be involved in this process by applying a bioinformatics approach to analyse large quantities of single cell RNA and small RNA sequencing data across multiple studies. We identified the Dlk1-Dio3 locus, a large cluster containing approximately 55 miRNA genes and many other non-coding RNAs, as a potential driver of cell states in glioblastoma. These miRNAs have potential utility as biomarkers and implicate a novel set of targets for therapeutic research.

Author: Mr Chris Smith, University of Technology Sydney

A bioinformatic filter that identifies the microRNA targetome in planta.

MicroRNAs (miRNAs) are 20-22nt long non-coding RNAs involved in the regulation of many important biological processes. In plants, efficient gene silencing requires high complementarity between the miRNA and its target mRNA, and this has enabled relatively simple complementary-based bioinformatic prediction of miRNA targets. However, from the many predicted targets, very few are experimentally verified, suggesting the prediction of many false positives and that a filter is needed to identify in planta miRNA targets. To address this, I built the Major MicroRNA Target Filter (MMTF), a computational pipeline based on degradome analysis which sequences all uncapped mRNA including miRNA cleavage products. Degradome analysis enables the detection of degradome signatures, that are placed into different categories (1-4) based on the frequency at which a miRNA cleavage product is detected. The MMTF analyses publicly available degradome data that constitutes multiple degradome experiments in multiple diverse species. Based on the strength and frequency of a degradome signature, a “Category Score” can be assigned for all predicted targets. This enables ranking of targets as either “Major” or “Minor” and has enabled the rapid identification of a more accurate plant miRNA “targetome” for the first time.

Author: Ms Gigi Wong, Australian National University

Contig-level assembly of the Duboisia myoporoides genome

Duboisia is a genus of shrubs rich in alkaloid contents native to Australia, which has been used as medicine to aboriginal people. Commercial cultivation of began in Queensland in the late 60s as pharmaceutical products.

We began studying Duboisia} by sequencing the genome of three species Duboisia myoporoides, Duboisia leichhardtii and Duboisia hopwoodii. Only Duboisia myoporoides was successfully sequenced on both Illumina and PacBio platform and was subsequently assembled de novo. K-mer profile suggested a relatively large, highly homozygous genome (1.5Gbps and 0.752% heterogyzous). Three mainstream assembler were used for de novo assembly, e.g. Canu, Falcon and Mecat2. Three assemblies were assessed by continuity and genome completeness. N50 ranged from 1.186Mbps to 443Kbps and all three raw assembles are reasonably complete.

All key genes implicated in alkaloid production from other solanaceous plants were identified in the polished assemblies. Polished assemblies were aligned against each other to visualize relationship between contigs which will serve as references for downstream analysis.

Author: Mr Siyuan Wang,The University of Queensland

Dysregulation of RPE immunosuppression during ageing: a RNA-seq study

Purpose: The ageing of the retinal pigment epithelium (RPE), and loss of subretinal immune privilege are thought to be fundamental in the development of age-related macular degeneration (AMD). This study characterised the global transcriptomic changes associated with RPE immunosuppression during ageing.

Methods: RPE-choroidal tissues were isolated from eyes of C57BL/6J mice at 3 (young) and 22 months old (aged). RNA-seq was performed on Illumina NovaSeq 6000 System using total tissue RNA. Differentially expressed genes (DEGs) analysis and functional overrepresentation analysis of DEGs were conducted via edgeR and clusterProfiler packages respectively in R.

Results: 5967 DEGs (FDR<0.05) were identified between young and aged RPE-choroid, e.g. the most dysregulated immune-modulating Orm1. Immunosuppressive DEGs Tgfb3, Thbs1, Serpinf1 and Ctla2a were significantly downregulated in ageing. 17 significantly overrepresented Gene Ontology biological processes were linked to myeloid leukocyte chemotaxis and inflammatory response (q<0.05). DEGs involved in phagocytosis and inflammatory leukocyte migration, namely Fcerlg, Itgb2 and Syk, were also recognised in previous microarray ageing studies using RPE-choroid.

Conclusions: We confirmed an age-related loss of RPE immunosuppression. With Fcerlg, Itgb2 and Syk being consistently identified across different transcriptome studies, these genes may account for recruiting subretinal phagocytic leukocytes with age, and the immune dysregulation leading to AMD.

Author: Miss Josephine Wong, The University of Melbourne
Co Author/s: Dr Alice Brandli, Dr Matt Rutar,Dr Andrew Jobling, Professor Erica Fletcher, The University of Melbourne

The Shape of Phylogenies Under Phase-Type Distributed Times to Speciation and Extinction

We consider a macroevolutionary model for phylogenetic trees where times to speciation or extinction events are drawn from a Coxian phase-type (PH) distribution. In this poster, we show that different choices of parameters in our model lead to a range of tree shapes as measured by Aldous’ β statistic. Additionally, we show that tree balance is mainly controlled by speciation process while branch lengths are mostly controlled by extinction process.

Author: Mr Albert Christian Soewongsono, University of Tasmania
Co Author/s: Professor Barbara Holland, Associate Professor MaÅ‚gorzata O’Reilly

Systematic evaluation for metrics of gene expression variability in single-cell RNA sequencing data

During ageing, transcriptional noise has been shown to increase in multiple organs and tissues. Transcriptional noise is defined as the variability of gene expression, and this property reflects the heterogeneity that results from stochastic cell to cell variation. Although the concept of transcriptional noise is not new, different metrics are being used to measure this and it is unclear what the optimal approach is. With the advent of single-cell sequencing techniques, it is becoming possible to quantify how noise is distributed through the genome. The project focuses on understanding how to accurately model transcriptional noise as a regulatory property of the genome and its contribution to the fundamental feature of ageing.

To conduct a systematic evaluation, we selected 12 different metrics that commonly used in scRNA-seq studies. Performance of these metrics is tested with simulated and experimentally-derived datasets. We investigated the performance of these metrics against different data structures, stably expressed genes and other properties. Using a publicly available scRNA-seq dataset with multiple tissues and age groups for mice, we intend to investigate how transcriptional noise changes during ageing. Through the analysis, the goal is to understand how transcriptional noise impacts the regulatory processes that underlie ageing.

Author: Miss Huiwen Zheng, The University of Queensland

Discovery of Tissue-specific Gene Expression Patterns in CD8 T Cells by Single-cell RNA-seq

Immune checkpoint blockade (ICB) has been argued to be the basis of next-generation immunotherapy. Nevertheless, although the efficacy of ICB has been attributed to CD8 T cells, the underlying precise mechanism remains poorly understood. In recent years, with the clinical application of ICB, the distinct effects of ICB on the treatment of different cancer types and the tissue-specific complications of ICB (immune-related adverse events, irAEs) have been widely reported. In allusion to the association between tissues and ICB effects, in this research, we used single-cell technologies to profile murine tissue-infiltrating CD8 T cells obtained from 8 different tissues, including the small intestine, kidney, liver, lymph nodes, lung, PBMCs, spinal cord, and spleen. Through the comprehensive comparison and characterisation of these cells, our research proves cellular heterogeneity of tissue-infiltrating CD8 T cells, profiles their abundant tissue-specific gene expression patterns, and most importantly, identifies several tissue-specific-infiltrating CD8 T cell subpopulations in the liver and kidney. The trajectory analysis further reveals the distinct differentiation statuses of infiltrating CD8 T cells across tissues and identified subpopulations. These discoveries have important implications for illustrating the underlying association between tissues and ICB, including the relationship between tissue-specific irAEs and ICB effects in different cancer types.

Author: Miss Ying Zheng, Australian National University
Co Author/s: Dr Di Yu,The University of Queensland, Dr Jiayu Wen, Australian National University

Image-based Predictive Modelling for the Characterisation of Cellular Senescence

Senescence is a cellular stress response that is hypothesised to act as the causal nexus of ageing through its extensive pro-inflammatory secretory phenotype. Cells entering senescence undergo extensive morphological, transcriptional and metabolic changes, which vary significantly by cell-type, inducing stressor, and time since induction. Due to its dynamic and heterogeneous nature, a specific and universal biomarker of senescence remains unidentified, presenting a major obstacle in the research and development of therapeutics.
Here we present the ongoing development of an image-based predictive model for senescence using deep convolutional neural networks and high-content microscopy of ageing MSCs, and their subsequently derived osteocytes, adipocytes and chondrocytes. Images are corrected, segmented and labelled in a pre-processing pipeline developed in Python 3, to be fed as training data to a supervised learning algorithm. Through post-hoc interpretability and visualisation methods, the final model can be analysed to provide meaningful insight into the complex relationships and features that phenotypically characterise the sub-types and transitional states of senescence. This will facilitate the identification of novel, robust biomarkers for improved therapeutic targeting. The models predictive classification of senescence from brightfield images alone will also make it a practical tool for assessing population quality in research and clinical settings.
.

Author: Ms Ebony Watson
Affilliation/University: Monday 2 December
Co Author/s: Dr Atefeh Taherian Fard, Dr Farhad Soheilmoghaddam,Professor Ernst Wolvetang,Associate Professor Jessica Mar
Co Author/s Affiliation/University: The University of Queensland,University of Southern Queensland