Clonal hematopoiesis of indeterminate potential (CHIP) is an age-associated condition linked to chronic inflammation and an increased risk of cardiovascular diseases and hematological malignancies. People with HIV (PWH) exhibit a higher prevalence of CHIP than the general population, but the mechanisms underlying this association remain unclear. In particular, it is unknown whether the excess burden of CHIP reflects earlier emergence of mutant clones, altered clonal expansion dynamics, or differences in selective pressures acting on hematopoietic stem cells. We reconstructed longitudinal trajectories of CHIP variant allele frequency (VAF) in 52 PWH using serial peripheral blood samples spanning up to 25 years from the Swiss HIV Cohort Study. We used spline-based modelling to estimate clone size and growth dynamics, and dynamic time warping to identify common trajectory patterns. Associations between clonal dynamics and longitudinal immune parameters were assessed using linear mixed-effects models. Trajectories in PWH were compared with publicly available longitudinal CHIP data from the SardiNIA population cohort. We identified heterogeneous clonal dynamics consistent with known gene-specific fitness patterns. Larger clone size was associated with lower CD4 T-cell count and lower CD4/CD8 ratio. Compared with the general population cohort, PWH showed higher VAF across the observed age range and steeper early trajectory increases, while long-term expansion rates were broadly similar. Greater variability in clonal dynamics among PWH suggests a stronger contribution of host environmental factors to clonal fitness. These findings support a model in which HIV-associated immune dysregulation alters the hematopoietic fitness landscape, contributing to earlier detectable clonal expansion and increased burden of CHIP in PWH.

This chapter reflects on research and teaching experiences - reading, interpretation and design - conducted in recent years, where the questions raised by the idea of the territory-subject and their controversies have emerged regularly and with severity, without ever retreating. It is as if it were no longer possible to go backwards but only to advance, even with difficulty, within the radical recontextualisation of the territory-subject and territorial subjectivation. In this approach, I include all the indeterminacies and problematic issues contained within the process, and that emerge with the ambiguity of the very term “subject” (sujectum-subjectus, as Étienne Balibar recalls) (Balibar, 2012). At the same time, the subject is individual, anthropological-political, but also subjected and subjugated, placed underneath. It is relational and not abstract (Pelluchon, 2020), a material, textured and bodily relationship, which exchanges with the other subjects.

In this work we introduce the automatically generated dataset in BLiMP-IT, a novel benchmark for evaluating Italian language models based on minimal pairs (i.e. sentence pairs that differ only in a critical morphosyntactic aspect). Drawing inspiration from the success of BLiMP for English, BLiMP-IT combines and adapts several existing resources—including COnVERSA, AcCompl-it, and BLiMP—to construct a high-quality evaluation dataset for Italian. We present an automatic methodology for generating the evaluation’s items by leveraging a large Italian corpus for lexicon extraction, POS tagging, and animacy annotations. Our approach not only ensures coverage of diverse morphosyntactic phenomena (e.g., agreement and inflection, verb class, non-local dependencies) but also scales the creation of minimal pairs to automatically expand the items for the evaluation benchmark. BLiMP-IT demonstrates that an automated pipeline for generating minimal pairs to evaluate LMs is both feasible and effective, ensuring comprehensive coverage of diverse morphosyntactic phenomena in Italian while reducing reliance on manual annotation.

Low photon extraction efficiency limits the optical spin readout fidelity in quantum sensing technologies in diamond. We demonstrate a potential spin-photon interface for enhancing the infrared readout of NV centers using a fiber-coupled microcavity.

We demonstrate the first high-speed lithium-tantalate-on-insulator (LTOI) Mach-Zehnder modulator (MZM) operating in the telecommunication O-band around 1310 nm. With PAM4, PAM6, and PAM8 signaling we achieved single-carrier line rates of 480 Gbit/s.

Peptide therapeutics have emerged as a powerful class of drugs capable of addressing challenging targets, particularly proteins involved in large and flat protein-protein interactions (PPIs). Despite their high selectivity and low toxicity, peptide- and protein-based drugs often suffer from limited stability, poor cell permeability, and low oral bioavailability. The introduction of non-natural amino acids and site-selective chemical modifications represents an effective strategy to overcome these limitations and to expand the chemical space accessible to peptide and protein therapeutics. Furthermore, site-selective protein modification provides valuable tools for investigating protein structure and function. This thesis focuses on the development of new synthetic methodologies for the generation of non-natural amino acids and for the site-selective modification of proteins. The first part of this work describes the development of a mild approach for the alpha-functionalization of glycine derivatives, enabling access to various non-natural amino acids. Using imine-based glycine surrogates bearing a leaving group on nitrogen, a broad range of nucleophiles, including sul-phur-, oxygen-, nitrogen-, and carbon-based nucleophiles, were successfully introduced in moderate to excellent yields. This methodology provides a practical route to alpha-functionalized amino acid derivatives and complements existing approaches that are limited to alkyl and aryl substitutions. The second part of this thesis addresses the challenge of achieving site-selective protein modification. Building on the reactivity of hypervalent iodine reagents, novel ethynylbenziodoxolone (EBX)-peptide conjugates bearing peptide ligands were developed for the selective functionalization of cysteine, tyrosine and tryptophan residues under mild conditions on purified proteins. Functionalization of Cys434 in the KELCH domain of KEAP1 proceeds either through direct thiol addition to the EBX reagent, forming an S-VBX adduct, or via alkyne transfer. In contrast, functionalization of tyrosine residues in the KEAP1 C434Y mutant and in the oncoprotein beta-catenin occurs exclusively through phenol addition to the EBX reagent. The peptide component serves as a recognition element, allowing targeting of specific regions on protein surfaces via non-covalent interactions between the protein and the ligand. Furthermore, the introduced functional handles enable subsequent derivatization through bioorthogonal azide-alkyne cycloaddition or palladium-mediated cross-coupling reactions. Overall, this work provides new tools for the synthesis of non-natural amino acids and for the precise chemical modification of proteins, contributing to the development of advanced strategies for peptide and protein functionalization in chemical biology and drug discovery.

The idea of extending human abilities by controlling robotic interfaces, such as extra fingers or arms, or by interacting with computers through neural signals, has long captured human imagination. Recent advances in neuroscience, robotics, and engineering have progressively transformed these concepts from science fiction into a concrete and rapidly emerging field of research known as human motor augmentation. One of the central challenges in this domain lies in identifying control policies that allow users to voluntarily and reliably operate external interfaces while remaining independent from, and non disruptive to, other physiological functions.

A prevailing perspective suggests that a promising solution may lie in exploiting the intrinsic redundancies of the human body, namely the fact that motor behavior is supported by more muscles, joints, and neurons than are strictly necessary to generate movement. This thesis investigates how such redundancies can be identified, evaluated, and leveraged to design control strategies for augmented motor control. Focusing on the problem of providing users with control over a third robotic arm, we first propose a human machine interface based on the voluntary modulation of the diaphragm as a kinematic null space control strategy. We then extend this approach to multi degree of freedom control by combining kinematic and muscular null space signals, specifically by integrating diaphragm modulation with contraction of the auricular muscles, vestigial muscles that represent an attractive and minimally interfering control source for augmentative devices.

Finally, this thesis introduces a novel framework grounded in recent advances in neural population dynamics and neural geometry. Using mesoscale neural signals recorded with electrocorticography, we demonstrate the possibility of identifying low dimensional neural spaces that selectively capture neural variance associated with motor imagery while remaining orthogonal to execution related dimensions. We propose that such imagery specific neural dimensions can be exploited to control external interfaces concurrently with overt movement while minimizing interference.

Overall, this thesis presents complementary kinematic, muscular, and neural strategies for motor augmentation, with a unifying emphasis on minimizing interference with concurrent biological functions. In doing so, it contributes both methodological frameworks and empirical evidence toward the development of robust and scalable augmentative technologies.

Active clays, particularly sodium smectite-rich bentonites, play a central role in a range of environmental and geotechnical engineering applications, most notably as sealing materials in landfills and deep geological repositories. Their suitability stems from their unique hydro-mechanical characteristics: exceptionally high swelling capacity, very low hydraulic conductivity, and a strong ability to retain water through both capillary and adsorption mechanisms. These properties arise from the mineralogical features of smectite, contributing to swelling, permeability reduction, and mechanical stability. Water in clays exists primarily in two forms: capillary water, retained in pore spaces, and adsorbed water, bound to mineral surfaces. Despite their distinct roles in controlling both hydraulic and mechanical responses, the current understanding and quantification of these mechanisms remain incomplete. This thesis advances the understanding, quantification, and modelling of hydro-chemo-mechanical behaviour in active clays, with a focus on sodium bentonite. The work addresses several interlinked gaps: (i) the lack of a robust macroscopic method to separate and quantify capillary and adsorbed water in compacted specimens without disturbing fabric; (ii) the absence of a composition-aware framework for comparing and predicting swelling pressure under varying salinity; (iii) the poor characterisation of water retention at lower dry densities and under saline wettingâ drying conditions, including the definition of â trueâ saturation when interlayer water density differs from bulk water; (iv) experimental uncertainty around saturation itself â saturation protocols for bentonite are difficult to validate using conventional total pressure and pore water pressure criteria because swelling pressure affects stress measurements, leaving full saturation and its confirmation insufficiently understood; and (v) the inconsistent use of â adsorbed waterâ across disciplines and its unclear geomechanical relevance. A central contribution is a thermogravimetric analysis (TGA)-based experimental protocol designed to distinguish and quantify adsorbed water in undisturbed specimens across different hydro-mechanical states, enabling physically meaningful partitioning between adsorbed and capillary water and supporting improved retention curves. Building on this, the thesis develops a semi-empirical predictive framework of swelling pressure that links mineralogical variations and salinity effects â supported by interlayer-spacing evidence â enabling more consistent cross-material interpretation. An experimental campaign is then further developed to investigate how salinity and compaction influence retention, hysteresis, saturation evolution, and adsorption mechanisms under controlled boundary conditions. Further, this thesis consolidates the experimental practice needed to obtain reliable hydro-mechanical data in highly expansive clays, detailing saturation and constant-volume testing challenges, typical sources of error, and methodological recommendations that improve reproducibility and interpretation. Finally, the thesis synthesises adsorption-scale physico-chemical mechanisms into an integrated conceptual definition tailored to geomechanics, establishing a basis for refining water retention formulations and for future developments in effective stress and constitutive modelling of active clays.

Hydraulic stimulation is a key method for creating permeable pathways in deep geothermal systems, where natural fracture networks are often too tight to allow economic fluid circulation. The effectiveness of stimulation depends on the ability to enhance permeability through shear slip on pre-existing discontinuities, while field operations are challenged by complex fluid-fault interactions and the frequent occurrence of injection-induced seismicity. A mechanistic understanding of how aseismic slip propagates and modifies permeability is therefore essential for designing effective and safe stimulation strategies.

This thesis investigates fluid-driven aseismic rupture with particular emphasis on shear-induced dilatancy, permeability evolution, and the role of the stress field. We begin with a semi-analytical treatment of rupture growth under a linear stress gradient, the first problem addressed in this work. This analysis provides fundamental insight into how in-situ stress variations influence the extent and asymmetry of fluid-driven slip.

We then employ a recently developed hydro-mechanical solver that models a poroelastoplastic interface with Mohr-Coulomb strength, slip-weakening friction, non-associated dilatancy, nonlinear contact response, and aperture-dependent permeability. After verifying the solver against semi-analytical benchmarks, we apply it to the Basel-1 geothermal stimulation. Using the recorded injection schedule together with geological and stress constraints, we reproduce both the wellhead pressure evolution and the migration of the dominant microseismic cluster. The analysis highlights how permeability increase due to dilation, nonlinear contact compliance, and variable permeability are required to reconcile observed pressure drops and rupture expansion.

Finally, we extend the theoretical understanding of fluid-driven aseismic slip by embedding dilatancy-controlled permeability evolution into a two-dimensional plane-strain formulation. The analysis identifies a new early-time diffuse-dilatant regime, arising from shear-induced dilatancy documented in laboratory experiments and rupture theory, in which rupture propagation is slowed, overpressures are depressed near the injection point, and self-similarity is lost compared to constant-permeability theory. A transition occurs once slip reaches the critical dilatancy distance, leading to faster growth and, in critically stressed cases, undrained pore-pressure drops. At late times, dilatancy localizes into a narrow process zone, and the solution converges toward the self-similar form predicted by abrupt-dilatancy theory. This framework clarifies the regimes where dilatant hardening is most influential and the well-pressure signatures by which it may be detected in the field.

This doctoral research provides a quantitative and conceptual framework for interpreting hydraulic stimulation data and for designing injection protocols that promote stable hydro-shearing and sustained permeability enhancement. By extending classical fluid-driven aseismic slip theory to include depth-dependent in-situ stresses and slip-induced permeability evolution, it clarifies when and why rupture growth becomes asymmetric, how dilatant hardening imprints diagnostic well-pressure transients, and how these effects can be reconciled with field observations such as Basel-1.

Spinal cord injury (SCI) causes persistent sensorimotor and autonomic deficits that severely impair quality of life and impose a substantial societal burden. Although limited spontaneous recovery can occur after injury, this recovery is typically incomplete. Neuromodulation strategies have emerged as promising approaches to enhance recovery, with epidural electrical stimulation (EES) demonstrating remarkable clinical efficacy. More recently, transcutaneous spinal cord stimulation (tSCS) has gained attention as a non-invasive alternative capable of improving motor function after SCI. Despite encouraging clinical outcomes, the biological mechanisms underlying tSCS-mediated recovery remain poorly understood. In this thesis, we developed a comprehensive preclinical framework to dissect the mechanisms of action of tSCS and to identify the neural circuits mediating recovery. We engineered a scalable, wearable stimulation device optimized for rodents, integrating novel electrode materials into a modular design that ensures stable and reproducible stimulation. This interface was combined with a robotic neurorehabilitation platform enabling precise, quantitative assessment of upper-limb motor function during longitudinal rehabilitation. Using this platform, we demonstrate that a clinically inspired high-carrier-frequency stimulation protocol (ARCEX) reproduces key functional benefits observed in humans. We show that non-invasive stimulation accesses the spinal cord primarily through proprioceptive sensory afferents, which relay stimulation evoked signals to Vsx2-expressing spinal interneurons. Through electrophysiology, transcriptional profiling, viral tracing, and targeted ablations, we establish that both proprioceptive afferents and Vsx2- expressing interneurons are required for ARCEX-mediated recovery. We further demonstrate that stimulation protocols differ fundamentally in their recruitment of sensory pathways. In contrast to ARCEX, conventional stimulation paradigms preferentially engage cutaneous nociceptive afferents, increase discomfort and pain-related neuronal activation, and fail to promote long-term functional recovery. Standardized pain metrics, molecular profiling of dorsal root ganglia, and pharmacological silencing of skin afferents reveal that excessive nociceptive recruitment constrains the therapeutic window of tSCS and disrupts rehabilitation-induced recovery. Finally, we place these findings in the context of spontaneous recovery after cervical SCI. Using whole-brain mapping, transcriptomics, and circuit-specific manipulations, we identify a distributed recovery architecture centered on the parvicellular reticular nucleus (PARN) and spinal Vsx2-expressing interneurons. We show that ARCEX engages and refines this endogenous recovery network, thereby amplifying natural repair mechanisms. Together, this work establishes a mechanistic framework linking non-invasive spinal cord stimulation to circuit level plasticity and functional recovery, and provides a biological rationale for optimizing tSCS protocols that prioritize proprioceptive recruitment while minimizing discomfort.