70040 research outputs found
Sort by
Understanding RNA-Peptide Interactions that Facilitate Liquid-Liquid Phase Separation
No full textRNA-protein interactions underpin many of the vital processes in the cell, such as the regulation of RNA and protein synthesis. This is performed through binding events between highly structured proteins with defined secondary and tertiary structures, but in recent decades the importance of so-called intrinsically disordered proteins has come to light. Lacking secondary and tertiary structure, they are able to form membraneless organelles in the cell through non-covalent interactions with non-coding RNAs. This process is called liquid-liquid phase separation (LLPS), and the disruption of this process can lead to solid aggregate formation, a process implicated to be responsible for some neurodegenerative diseases. Therefore, it is imperative to understand what features of proteins and RNAs that enable them to undergo LLPS.
In this thesis, short RNA and peptide oligomers were analysed in a variety of ways to investigate their ability to interact with each other and undergo LLPS. By using these shorter oligomers, it was possible to ‘zero in’ better on which structural motifs were contributing, when compared to using whole proteins and RNAs in a biological context. Firstly, 1H NMR titrations and ITC were used to attempt to quantify the binding strength between simple RGG repeat peptides and homopentamer RNAs. These interactions were weak, and difficult to quantify, but it was determined they would interact with a Ka on the order of magnitude of 1000 M-1, with a slight preference for the adenine-rich RNA, A5.
Another limitation with a lot of the in vitro research into LLPS, is that removed from the context of the cell, the impact of macromolecular crowding is ignored. It was shown that a model cationic peptide showed preference for undergoing LLPS with purine rich (AG)5 RNA over pyrimidine rich (CU)5. However, under the crowding condition of poly(ethylene glycol) (PEG), this preference was lost with longer PEG polymers (≥1500 g mol-1). This highlights the importance of the macromolecular crowding effect on LLPS, and that scientists should be mindful of the character and size of crowders they use.
All the peptides synthesised in this Thesis were based on the arginine-glycine-glycine repeat, ubiquitous in the proteins which form membraneless organelles in the cell. This is believed to be due to arginine’s positive charge, as it can bind to RNA via electrostatic interactions with the phosphate backbone. By changing the amino acids in the peptide to lysine, the other positively charged amino acid, it was determined that other features of arginine along with the charge must contribute to its ability to facilitate LLPS, such as stronger hydrogen bonding and ability to partake in cation-π and π-π interactions.
Finally, a redox-responsive RNA containing system that could undergo LLPS was developed through the incorporation of the thiol-containing amino acid cysteine. This demonstrates that membraneless organelles can be sensitive to redox chemistry, a process which has been suggested to occur under cellular stress
Extracting Non-Linear Representations for Concept Control
No full textLarge language models have received much attention in recent years, but despite their ability to generate coherent responses for prompts regarding a wide range of disciplines and applications, they remain prone to generating biased, offensive or hallucinatory text. This may lead to widespread consequences such as the spread of misinformation or discrimination, users following harmful advice and reduced trust in artificial intelligence. Hence, safety is one of the most
important open problems with generative models.
Representation engineering is a recent approach that has shown promising results with alleviating this issue. Existing representation engineering methods extract vectors representing the direction of concepts such as truth or toxicity from a language model using as few as two prompts. By adding increasing multiples of a ‘truth’ vector to the activations, the truthfulness of a model can be continuously increased, providing a simple and effective means of controlling the behaviour of a model, without the need for external datasets or the precomputation associated with fine-tuning.
However, it is unlikely that these simple, linear representations are able to approximate a complex model’s ground-truth representation of such high-level, vague and nuanced concepts to a high degree of accuracy. This may explain why the quality of responses degrades significantly as the behaviour of the model is increasingly controlled by a linear representation.
This thesis investigates the potential of non-linear representations to form better local approximations of concepts through the proposal of three non-linear representation methods: Namely, TaylorRep, CurveRep and SurfaceRep. TaylorRep generalises existing linear methods by representing concepts with Taylor approximations. Whilst this representation controls activations directly in the model’s latent space, the two other methods represent concepts in an intermediate,
low-dimensional space using a polynomial and radial basis function network, respectively. Two important components of these methods are an algorithm to collect model activations required to fit non-linear representations and a method to control the model by mapping from low-dimensional representations to high-dimensional activations. Additionally, this thesis is the first to adapt representation engineering methods to image generation model control. We bridge this gap
through a comparison to the implementation of existing image generation control methods.
In addition to qualitative analysis of text and images generated by models controlled by these representations, we evaluate the methods on the BBQ, RealToxicityPrompts and TruthfulQA benchmarks. Although linear methods maintain the advantages of speed and simplicity, TaylorRep outperforms existing methods on both reduction and control of toxicity. Furthermore, the projection-based CurveRep and SurfaceRep methods have the advantage that their low-dimensional representations are readily visualisable
Quasi-Constitutionality and Indigenous-State Treaties in Australia
No full textThe Statewide Treaty negotiated between the First Peoples Assembly of Victoria and the Victorian government is momentous. However, the Statewide Treaty Bill 2025 (Vic), which implements the agreement, is an ordinary piece of legislation subject to the ordinary rules of statutory interpretation and implied repeal. Drawing on jurisprudence from Canada, the United Kingdom and scholarship from the United States, we argue that the anticipated Statewide Treaty Act should properly be understood as ‘quasi-constitutional’ because it: (i) shapes the character of Victoria in moral and legal ways, (ii) is understood within the broader community as a statute of great importance, and (iii) regulates the institutions and powers of the state. This designation has important legal and expressive implications which supports the status and prominence of Indigenous-State treaties
Acute blood pressure responses to plank and wall sit isometric exercise in adults
No full textOBJECTIVES: Isometric exercise is an accessible, time efficient intervention for reducing blood pressure (BP). However, guideline recommendations for its use remain sparse due to safety concerns. This study aimed to determine the proportion of participants whose BP exceeded predefined limits during isometric exercise. Secondary aims were to examine differences in BP responses between sexes and across exercises. METHODS: Participants attended the laboratory for two sessions: screening and familiarisation, and exercise testing. In session two, participants performed the plank and wall sit in a randomised order while their BP was continuously monitored noninvasively using the Finapres NOVA. Each exercise was performed until volitional fatigue or until reaching the American College of Sports Medicine limits of SBP (>250 mmHg) or DBP (>115 mmHg). RESULTS: Sixty-two apparently healthy participants completed the study (age 26.7 ± 8.2 years, 52% women, resting BP 121 ± 8/77 ± 6 mmHg). Almost all participants were stopped [plank: 97% (95% confidence interval, 95% CI 89-99); wall sit: 98% (95% CI 91-100)] due to DBP more than 115 mmHg after an average of approximately 50 s of exercise. All women exceeded the DBP limit [plank: 100% (95% CI 89-100); wall sit: 100% (95% CI 89-100)] compared to most males [plank: 93% (95% CI 78-99); wall sit: 97% (95% CI 83-99)]. No participants reached the SBP limit. No adverse events were reported. CONCLUSION: Common isometric exercises cause a marked increase in BP, particularly DBP. Despite this, no adverse events occurred. Future studies should examine the dose-response of different isometric exercises in higher risk populations to better inform its clinical suitability
Mental health service use during childhood and adolescence: An Australian longitudinal population cohort study.
No full textCombination of two or more drugs in the same syringe among a national sample of people who regularly inject drugs in Australia, 2025
No full textThe Illicit Drug Reporting System (IDRS) has been conducted annually since 2000. The IDRS is a national drug monitoring system, designed to identify and track emerging trends of local and national concern in illicit drug use and markets. The IDRS forms part of the broader Drug Trends program. This bulletin examines the percentage of participants who combined two or more drugs in the same syringe in the month prior to interview, how many times the drugs were combined in the same syringe, the specific drugs combined and sociodemographic and health factors associated with participants who reported combining two or more drugs in the same syringe
Unraveling the role of gastric clots in proteolysis and lipolysis kinetics: Insights from a novel in vitro dynamic infant digestion system for animal and plant-based formulas
No full textUnderstanding how different protein sources influence digestion is critical for optimizing infant formula design. While gastric clot formation is known to modulate digestion kinetics in infants, the specific impact of protein origin on this process remains poorly understood. In this study, we present a novel dynamic in vitro infant stomach-duodenum digestion system (DIS-II) that, for the first time, reproduces key physiological features of the gastrointestinal tract of young children, including anatomical configuration, pH gradients, peristaltic motion, and gastric emptying behavior. Using DIS-II system, we investigated the digestion dynamics of four commercial powdered formulas: concentrated whole cow milk (CWCM), skimmed cow milk (CM), goat milk (GM), and soy milk (SM). After 30 min of gastric digestion (from 142 g of reconstituted milk), visible clot formation occurred in CWCM (40.3 g), CM (28.8 g), and GM (14.3 g), while SM showed no visible clotting. Clot mass declined over time due to mechanical disruption, gastric emptying, and proteolysis. Rheological measurements indicated that CWCM formed the most rigid and viscous clots, exhibiting higher shear modulus and viscosity than those from CM and GM. Despite similar half-emptying times across all formulas (71–77 min), GM and SM exhibited faster gastric emptying at later stages (>90 min). Proteolysis was most efficient in GM (72.0 %) and CM (71.0 %), followed by SM (61.1 %) and CWCM (54.6 %). Lipolysis, assessed via free fatty acid (FFA) release, followed the trend SM (0.39 g/g fat) > GM (0.32 g/g fat) > CM (0.28 g/g fat) > CWCM (0.25 g/g fat), likely due to the degree of lipid entrapment within the clots, which limited enzyme-lipid interaction. This study is the first to comprehensively model the dynamic interplay between gastric clot formation and digestive kinetics using an infant-specific in vitro system. These findings emphasize the critical role of gastric clot formation in regulating digestive kinetics and highlight the need to further explore how protein sources affect nutrient absorption in infants
Initial Teacher Education: Addressing the Needs and Perceived Challenges of First- and Second-Career Pre-Service Teachers
No full textLearning to teach represents a complex process of professional and personal transformation that involves constructing new professional identities, negotiating prior experiences, and engaging with the epistemologies of teaching. For both first- and second-career entrants, ITE is not only a pathway into employment but a significant phase of identity (re)formation and learning within the higher education context. Understanding the needs and challenges that pre-service teachers articulate during this phase therefore provides critical insight into how teacher education functions as a space of learning and belonging. Despite this, empirical research addressing this issue remains limited. Accordingly, the present study sought to address this gap by exploring the differences in the needs and challenges of first- and second-career pre-service teachers. Analysis of qualitative data collected from 123 Australian first- and second-career pre-service teachers revealed that both groups identified a strong need for more authentic, practice-based learning and sustained mentorship to bridge the persistent gap between theory and classroom practice. Despite shared concerns around behaviour management, assessment, and wellbeing, second-career entrants emphasised challenges related to transferring prior professional identities, navigating school cultures, and balancing study with family and financial responsibilities. These findings highlight the importance of differentiated, career-responsive approaches in initial teacher education that recognise diverse life experiences as central to professional learning and teacher identity formation
Quantifying changes to solute segregation behaviour at interfaces in additively manufactured Ti-6Al-4V
No full textElectron powder bed fusion has emerged as a promising additive manufacturing process for producing Ti-6Al-4V components with complex geometries. Although prior work has examined microstructural evolution phenomena including how thermal history influences the prevalence of specific α/α grain boundary misorientation intervariants governed by the Burgers orientation relationship between the α and β phases, the solute segregation behaviour at these interfaces remains poorly understood. This study employs correlative transmission Kikuchi diffraction and atom probe tomography to quantify the Gibbsian interfacial excess of V, Fe, and Al across multiple interfaces in electron powder bed fusion produced Ti-6Al-4V. V and Fe exhibited enrichment, while Al was depleted, with V segregation generally increasing with boundary misorientation. Spatial variations in segregation were observed across interface planes, although no consistent trend with boundary curvature was identified. First-principles calculations were also performed on model α/α grain boundaries to evaluate segregation energetics, confirming a thermodynamic preference for V enrichment and Al depletion. The agreement between experimental observations and DFT predictions highlights the influence of local defect structures and atomic-scale interactions on segregation behaviour. These findings improve understanding of grain boundary chemistry in additively manufactured Ti alloys and provide a basis for future grain boundary engineering approaches aimed at tailoring interfacial properties
