Scientific Publications
Co-assembled supramolecular hydrogels: nano-IR sheds light on tripeptide assemblies
Abstract Supramolecular hydrogels composed of self-assembling short peptides are gaining momentum for enzyme mimicry. In particular, multicomponent systems that feature similar peptides with a self-assembling motif (e.g., Phe–Phe) and catalytic residues (e.g., His, Asp) offer a convenient approach to organize in space, functional residues that typically occur at enzymatic active sites. However, characterisation of these systems, and especially understanding whether the different peptides co-assemble or self-sort, is not trivial. In this work, we study two-component hydrogels composed of similar tripeptides and describe how nano-IR can reveal important details of their packing, thus demonstrating it to be a useful technique to characterise multicomponent, nanostructured gels. Authors Paola Alletto, Ana M. Garcia, Federica Piccirilli, Silvia Marchesan Journal Faraday Discussions Publication date 21/02/2025 Consult the publication
Interpreting and Steering Protein Language Models through Sparse Autoencoders
Abstract The rapid advancements in transformer-based language models have revolutionized natural language processing, yet understanding the internal mechanisms of these models remains a significant challenge. This paper explores the application of sparse autoencoders (SAE) to interpret the internal representations of protein language models, specifically focusing on the ESM-2 8M parameter model. By performing a statistical analysis on each latent component’s relevance to distinct protein annotations, we identify potential interpretations linked to various protein characteristics, including transmembrane regions, binding sites, and specialized motifs. We then leverage these insights to guide sequence generation, shortlisting the relevant latent components that can steer the model towards desired targets such as zinc finger domains. This work contributes to the emerging field of mechanistic interpretability in biological sequence models, offering new perspectives on model steering for sequence design. Authors Edith Natialia Villegas Garcia, Alessio Ansuini Journal Workshop on Generative and Experimental Perspectives for Biomolecular Design @ ICLR 2025 Publication Date 13/02/2025 Consult the publication
Striped Twisted State in the Orientational Epitaxy on Quasicrystals
Abstract The optimal “twisted” geometry of a crystalline layer on a crystal has long been known, but that on a quasicrystal is still unknown and open. We predict analytically that the layer equilibrium configuration will generally exhibit a nonzero misfit angle. The theory perfectly agrees with numerical optimization of a colloid monolayer on a quasiperiodic decagonal optical lattice. Strikingly different from crystal-on-crystal epitaxy, the structure of the novel emerging twisted state exhibits an unexpected stripe pattern. Its high anisotropy should reflect on the tribomechanical properties of this unconventional interface. Authors Nicola Manini, Mario Forzanini, Sebastiano Pagano, Marco Bellagente, Martino Colombo, Dario Bertazioli, Tommaso Salvalaggio, Andrea Vanossi, Davide Vanossi, Emanuele Panizon, Erio Tosatti, Giuseppe E. Santoro Journal Physical Review Letters Publication Date 12/02/2025 Consult the publication
Towards the FAIRification of Scanning Tunneling Microscopy Images
Abstract In this paper, we describe the data management practices and services developed for making FAIR compliant a scientific archive of Scanning Tunneling Microscopy (STM) images. As a first step, we extracted the instrument metadata of each image of the dataset to create a structured database. We then enriched these metadata with information on the structure and composition of the surface by means of a pipeline that leverages human annotation, machine learning techniques, and instrument metadata filtering. To visually explore both images and metadata, as well as to improve the accessibility and usability of the dataset, we developed “STM explorer” as a web service integrated within the Trieste Advanced Data services (TriDAS) website. On top of these data services and tools, we propose an implementation of the W3C PROV standard to describe provenance metadata of STM images. Authors Tommaso Rodani, Elda Osmenaj, Alberto Cazzaniga, Mirco Panighel, Cristina Africh, Stefano Cozzini Journal Data Intelligence Publication Date 07/02/2025 Consult the publication
Emergence of SARS-CoV-2 subgenomic RNAs that enhance viral fitness and immune evasion
Abstract Coronaviruses express their structural and accessory genes via a set of subgenomic RNAs, whose synthesis is directed by transcription regulatory sequences (TRSs) in the 5′ genomic leader and upstream of each body open reading frame. In SARS-CoV-2, the TRS has the consensus AAACGAAC; upon searching for emergence of this motif in the global SARS-CoV-2 sequences, we find that it evolves frequently, especially in the 3′ end of the genome. We show well-supported examples upstream of the Spike gene—within the nsp16 coding region of ORF1b—which is expressed during human infection, and upstream of the canonical Envelope gene TRS, both of which have evolved convergently in multiple lineages. The most frequent neo-TRS is within the coding region of the Nucleocapsid gene, and is present in virtually all viruses from the B.1.1 lineage, including the variants of concern Alpha, Gamma, Omicron and descendants thereof. Here, we demonstrate that this TRS leads to the expression of a novel subgenomic mRNA encoding a truncated C-terminal portion of Nucleocapsid, which is an antagonist of type I interferon production and contributes to viral fitness during infection. We observe distinct phenotypes when the Nucleocapsid coding sequence is mutated compared to when the TRS alone is ablated. Our findings demonstrate that SARS-CoV-2 is undergoing evolutionary changes at the functional RNA level in addition to the amino acid level. Authors Harriet V Mears, George R Young, Theo Sanderson, Ruth Harvey, Jamie Barrett-Rodger, Rebecca Penn, Vanessa Cowton, Wilhelm Furnon, Giuditta De Lorenzo, Margaret Crawford, Daniel M Snell, Ashley S Fowler, Anob M Chakrabarti, Saira Hussain, Ciarán Gilbride, Edward Emmott, Katja Finsterbusch, Jakub Luptak, Thomas P Peacock, Jérôme Nicod, Arvind H Patel, Massimo Palmarini, Emma Wall, Bryan Williams, Sonia Gandhi, Charles Swanton, David LV Bauer Journal PLoS biology Publication Date 21/01/2025 Consult the publication
The Geometry of Tokens in Internal Representations of Large Language Models
Abstract We investigate the relationship between the geometry of token embeddings and their role in the next token prediction within transformer models. An important aspect of this connection uses the notion of empirical measure, which encodes the distribution of token point clouds across transformer layers and drives the evolution of token representations in the mean-field interacting picture. We use metrics such as intrinsic dimension, neighborhood overlap, and cosine similarity to observationally probe these empirical measures across layers. To validate our approach, we compare these metrics to a dataset where the tokens are shuffled, which disrupts the syntactic and semantic structure. Our findings reveal a correlation between the geometric properties of token embeddings and the crossentropy loss of next token predictions, implying that prompts with higher loss values have tokens represented in higher-dimensional spaces. Authors Karthik Viswanathan, Yuri Gardinazzi, Giada Panerai, Alberto Cazzaniga, Matteo Biagetti Journal Preprint Publication Date 17/01/2025 Consult the publication
Intrinsic Dimension Correlation: uncovering nonlinear connections in multimodal representations
Abstract To gain insight into the mechanisms behind machine learning methods, it is crucial to establish connections among the features describing data points. However, these correlations often exhibit a high-dimensional and strongly nonlinear nature, which makes them challenging to detect using standard methods. This paper exploits the entanglement between intrinsic dimensionality and correlation to propose a metric that quantifies the (potentially nonlinear) correlation between high-dimensional manifolds. We first validate our method on synthetic data in controlled environments, showcasing its advantages and drawbacks compared to existing techniques. Subsequently, we extend our analysis to large-scale applications in neural network representations. Specifically, we focus on latent representations of multimodal data, uncovering clear correlations between paired visual and textual embeddings, whereas existing methods struggle significantly in detecting similarity. Our results indicate the presence of highly nonlinear correlation patterns between latent manifolds. Authors Lorenzo Basile, Santiago Acevedo, Luca Bortolussi, Fabio Anselmi, Alex Rodriguez Journal International Conferenence of Learning Representations (ICLR) 2025 Publication Date 22/01/2025 Consult the publication
CO2 and Temperature Induced Switching of a Flexible Metal–Organic Framework with Surface-Mounted Nanoparticles
Abstract Within the material family of metal–organic frameworks (MOFs) the subclass of flexible MOFs (flexMOFs) has attracted great attention, showing structural flexibility as a response to external stimuli such as guest adsorption, temperature, and pressure. Hybrid composites like nanoparticle (NP) loaded flexible MOFs, which stand to potentially combine advantageous properties of both are yet largely unexplored. Here the synthesis of flexMOFs with surface mounted nanoparticles, e. g. NP@Zn2(BME-bdc)2dabco composites (NP = Pt and SiO2 nanoparticles, BME-bdc2− = 2,5-bismethoxyethoxy-1,4-benzenedicarboxylate, dabco = 1,4-diazabicyclo[2.2.2]octane) is reported, studying the impact of nanoparticles on the stimulus-responsiveness of a flexMOF. It is shown that CO2 physisorption triggered flexibility of the MOF is retained and reversible for all NP@flexMOF composites. Additionally, it is observed that NPs stabilize the large pore state of the MOF, slightly increasing and shifting the switching pressure window. This effect is also observed during temperature-induced switching but Pt@flexMOF composites partially lose long-range order during the reversion to their narrow pore state, while attached SiO2 NPs allow for a fully reversible transition. These findings suggest that the total exerted material strain triggering the switching is heavily dependent on NP size and the applied stimulus and that guest-induced switchability can be fully realized in NP@flexMOF hybrid materials. Authors Jan Berger, Stephanie Terruzzi, Hana Bunzen, Filippo Ballerini, Marco Vandone, Marcello Marelli, Luca Braglia, Roland A. Fischer, Valentina Colombo, Gregor Kieslich Journal Small Date 07/01/2025 Consult the paper
Innovation in rare earths recycling: a quantitative and qualitative analysis of patent data
Abstract Rare earth elements (REE) are currently essential enablers of the digital and decarbonization transition. However, their supply chain is highly concentrated and their extraction has a high environmental impact. Circular economy solutions could provide a double benefit, reducing supply risk for import-dependent countries and mitigating the impacts of REE mining. This article focuses on REE recycling and provides a comprehensive global overview of innovation dynamics in this sector using patent data. We propose a two-step patent search methodology to identify REE recycling patents, based on the OECD ENV-TECH classification for green technologies and keyword occurrence. We then develop a set of quantitative and qualitative metrics to explore innovation dynamics at the country, applicant and technology type level. China clearly emerges as the most attractive market for REE recycling patents and Chinese universities as the most active applicants worldwide. Conversely, patent applications in all other countries showed stagnating trends over the last decade. In particular, Europe has a lower number of both patent applications and patents developed compared to the US and Japan. However, patent quality indicators show a very different picture: US and Japanese applicants, who appear to be at the technological forefront, receive more citations and are more oriented towards protecting their inventions internationally. Our analysis therefore highlights the importance of considering both quantitative and qualitative patent metrics when examining innovation trends in REE recycling. We discuss the determinants of these observed phenomena, draw policy implications – particularly for REE import-dependent countries – and propose avenues for future research at the intersection of CRM, the circular economy, and innovation studies. Authors Riccardo Priore , Marco Compagnoni , Marinella Favot Journal Resources Policy Volume 102, March 2025, 105519 Date Available online 24/02/2025 Consult the paper
Ultrasensitive detection and quantification of bovine Deltapapillomavirus in the semen of healthy horses.
Abstract BPV1, BPV2, BPV13, and BPV14 are all genotypes of bovine delta papillomaviruses (δPV), of which the first three cause infections in horses and are associated with equine sarcoids. However, BPV14 infection has never been reported in equine species. In this study, we examined 58 fresh and thawed commercial semen samples from healthy stallions. In 34 (58.6%), bovine δPV DNA was detected and quantified using droplet digital polymerase chain reaction (ddPCR). Real time quantitative PCR (qPCR) was able to identify bovine δPV DNA in 5 samples (8.6%). Of the BPV-infected semen samples, 15 were positive for BPV2 (~ 44.1%) on ddPCR and 4 (~ 11.7%) on qPCR; 12 (~ 35.3%) for BPV14 on ddPCR and 1 (~ 3%) by qPCR; 4 (~ 11.7%) for BPV1 on ddPCR, whereas qPCR failed to reveal this infection; 3 (~ 8.8%) for BPV13 on ddPCR; and BPV13 infection was not detected by qPCR. Our study showed for the first time that BPV14 is an additional infectious agent potentially responsible for infection in horses, as its transcripts were detected and quantified in some semen samples. Large-scale BPV14 screening is necessary to provide substantial data on the molecular epidemiology for a better understanding of the geographical divergence of BPV14 prevalence in different areas and how widespread BPV14 is among equids. Authors Anna Cutarelli, Francesca De Falco, Francesco Serpe, Simona Izzo, Giovanna Fusco, Cornel Catoi, Sante Roperto Journal Scientific Report Publication date 04/01/2025 Consult the publication
Phenotypic evolution of SARS-CoV-2 spike during the COVID-19 pandemic
Abstract SARS-CoV-2 variants are mainly defined by mutations in their spike. It is therefore critical to understand how the evolutionary trajectories of spike affect virus phenotypes. So far, it has been challenging to comprehensively compare the many spikes that emerged during the pandemic in a single experimental platform. Here we generated a panel of recombinant viruses carrying different spike proteins from 27 variants circulating between 2020 and 2024 in the same genomic background. We then assessed several of their phenotypic traits both in vitro and in vivo. We found distinct phenotypic trajectories of spike among and between variants circulating before and after the emergence of Omicron variants. Spike of post-Omicron variants maintained enhanced tropism for the nasal epithelium and large airways but displayed, over time, several phenotypic traits typical of the pre-Omicron variants. Hence, spike with phenotypic features of both pre- and post-Omicron variants may continue to emerge in the future. Authors Wilhelm Furnon, Vanessa M Cowton, Giuditta De Lorenzo, Richard Orton, Vanessa Herder, Diego Cantoni, Georgios Ilia, Diogo Correa Mendonca, Karen Kerr, Jay Allan, Nicole Upfold, Gavin R Meehan, Siddharth Bakshi, Udeet Ranjan Das, Sergi Molina Arias, Marion McElwee, Sarah Little, Nicola Logan, Kirsty Kwok, Katherine Smollett, Brian J Willett, Ana Da Silva Filipe, David L Robertson, Joe Grove, Arvind H Patel, Massimo Palmarini Journal Nature Microbiology Publication Date 03/01/2025 Consult the publication
Limiting serine availability during tumor progression promotes muscle wasting in cancer cachexia
Abstract Cancer cachexia is a multifactorial syndrome characterized by a progressive loss of body weight occurring in about 80% of cancer patients, frequently representing the leading cause of death. Dietary intervention is emerging as a promising therapeutic strategy to counteract cancer-induced wasting. Serine is the second most-consumed amino acid (AA) by cancer cells and has emerged to be strictly necessary to preserve skeletal muscle structure and functionality. Here, we demonstrate that decreased serine availability during tumor progression promotes myotubes diameter reduction in vitro and induces muscle wasting in in vivo mice models. By investigating the metabolic crosstalk between colorectal cancer cells and muscle cells, we found that incubating myotubes with conditioned media from tumor cells relying on exogenous serine consumption triggers pronounced myotubes diameter reduction. Accordingly, culturing myotubes in a serine-free medium induces fibers width reduction and suppresses the activation of the AKT-mTORC1 pathway with consequent impairment in protein synthesis, increased protein degradation, and enhanced expression of the muscle atrophy-related genes Atrogin1 and MuRF1. In addition, serine-starved conditions affect myoblast differentiation and mitochondrial oxidative metabolism, finally inducing oxidative stress in myotubes. Consistently, serine dietary deprivation strongly strengthens cancer-associated weight loss and muscle atrophy in mice models. These findings uncover serine consumption by tumor cells as a previously undisclosed driver in cancer cachexia, opening new routes for possible therapeutic approaches. Authors Erica Pranzini, Livio Muccillo, Ilaria Nesi, Alice Santi, Caterina Mancini, Giulia Lori, Massimo Genovese, Tiziano Lottini, Giuseppina Comito, Anna Caselli, Annarosa Arcangeli, Lina Sabatino, Vittorio Colantuoni, Maria Letizia Taddei, Paolo Cirri, Paolo Paoli Journal Cell Death Discovery Publication date 21/12/2024 Consult the publication