Can we reconstruct evolutionary relationships without knowing the identity of the molecules involved? We developed a method combining mass spectrometry and Assembly Theory that does exactly this — inferring how species are related by analyzing the complexity structure of their molecular mixtures, without needing to sequence DNA or identify individual compounds. Applied to a range of biological and non-biological samples, the approach successfully recapitulates the tree of life, tracks bacterial lineages, and opens the door to studying evolutionary relationships in chemical domains where sequence-based methods are notably limited.

For life to emerge, chemistry had to become self-reproducing — but how likely is that to happen by chance? We explored this using the GARD model, a physicochemically rigorous simulation of self-replicating lipid networks. Self-reproducing compositions turn out to be rare in the space of possible chemistries, which would seem to make their spontaneous appearance unlikely. But we found a surprising saving grace: every self-reproducing state is a dynamic attractor of the catalytic network, meaning the system is naturally drawn toward these states over time. This dramatically increases the plausibility of life's emergence from disordered chemistry.

The conventional picture of the first protocells involves RNA or proteins enclosed in a lipid bilayer. But how did those biopolymers get there in the first place? We argue that self-reproducing lipid micelles came first — simpler structures capable of catalysis, growth, and compositional inheritance, requiring none of life's later molecular machinery. Drawing on experiments in micellar catalysis and our own computational modelling, we lay out a scenario in which lipid assemblies drove the earliest stages of molecular evolution, seeding the complexity that eventually gave rise to cellular life.

A living system is a selective chemical network that evolves from simple origins. To promote evolution, protocellular networks must exhibit sufficient selectivity in their molecular interactions, guiding molecular assembly. To test this, we conducted over 5600 organocatalytic reactions in two chemical systems, revealing that greater molecular assembly reduce average catalytic selectivity, highlighting a trade-off between exploration and selection. Both systems also showed a selectivity barrier as complexity increased, explaining the gap between living and non-living systems. This work provides insight into the kinetic foundations of selection in catalytic networks and the transition from chemistry to life.

RNA may be ancient, but was it the primal substrate of biology? Building on biochemical evidence for RNA ancestry, the case for RNA primacy relies on highly complex molecules whose spontaneous abiotic appearance is overwhelmingly implausible. The experimental evidence accumulated over decades has largely shown RNA's compatibility with prebiotic conditions — not its probability, necessity, or precedence in the vast combinatorics of prebiotic chemistry. Therefore, RNA's emergence demands a preceding protobiological stage, comprising collectively autocatalytic networks sustaining molecular evolution. Confronting the true combinatorial complexity of abiotic chemistry is essential for an unbiased account of life's origin.

For mixed lipid micelles to serve as protocells, they need to grow in a way that preserves their composition — otherwise, the chemical "memory" they carry would be lost with each generation. We used Molecular Dynamics simulations to watch this process unfold at the atomic scale, tracking how individual lipid monomers are incorporated into mixed micelles under different conditions. The simulations reveal that both the size and the chemical composition of a micelle significantly influence how fast it grows and which lipids it preferentially absorbs. These kinetic differences provide a plausible physical mechanism for compositional homeostasis — the ability of a micelle to reproduce its own chemical identity — lending concrete support to the idea that lipid assemblies could have carried heritable information before the advent of nucleic acids.

Despite decades of research, a substantial fraction of the yeast genome remains functionally uncharacterized — genes whose protein products exist but whose jobs in the cell are unknown. We built a systematic approach to change that: a database combining sequence and structural similarity predictions to nominate candidate enzymes for uncharacterized biochemical reactions. Applying this method, we identified and experimentally validated a previously unknown enzyme — Yhr202w, which we renamed Smn1 — as an extracellular AMP hydrolase in the NAD degradation pathway. Beyond the specific discovery, the method itself provides a reusable template for tackling the "dark proteome" of yeast and other organisms.

When we encounter a pathogen — or receive a vaccine — our immune system encodes a record of that event in the form of antibodies. Reading that record accurately and at scale has long required laborious single-antigen assays. We developed FLU-LISA, a high-throughput antigen microarray platform that simultaneously profiles antibody responses against many antigens in a single experiment, covering IgG, IgA, and IgM classes. Validated against influenza and SARS-CoV-2 panels in mice, chickens, and humans, FLU-LISA matches the sensitivity and specificity of traditional ELISA while processing hundreds of samples per day — and it works with dried blood spots, dramatically lowering the barrier for large-scale immune surveillance.

The earliest proteins are thought to have been rich in acidic amino acids, because the basic amino acids that make modern proteins fold stably were scarce in prebiotic chemistry. But highly acidic proteins don't fold — they repel themselves into disordered tangles. How, then, could early proteins have functioned? We tested one possible answer: polyamines, small positively charged molecules that are abundant in cells and were likely present in prebiotic environments, might have acted as primitive chemical chaperones. We engineered an ancestral 60-residue protein to be maximally acidic, confirmed it was disordered, and showed that polyamines restore folding at biologically plausible concentrations — with effectiveness scaling with the number of amine groups. The findings suggest polyamines may have been essential cofactors for early protein function, and could still play underappreciated roles in protein regulation today.

Saturn's moon Enceladus vents plumes of material from its subglacial ocean into space — and those plumes, it turns out, contain complex organic molecules. We analyzed these findings through the lens of origin-of-life chemistry and found a striking parallel: the monomers detected resemble simple lipids, while the larger polymers are chemically similar to terrestrial kerogens and organic compounds found in carbonaceous meteorites. We propose that these polymers could break down under Enceladus' conditions to yield micelle-forming amphiphiles — exactly the kind of material that, on the Lipid World hypothesis, could have seeded the origin of life. Enceladus may thus offer humanity its first direct window into a plausible prebiotic chemistry operating right now, providing an unprecedented opportunity to test competing origin-of-life models empirically.