Modal Geometry Governs Proteoform Dynamics arxiv.org/abs/2508.20004

It is remarkable how we are able to improve the performance and capabilities of our DIA-NN proteomics software at an ever-increasing pace. Sneak peek at the upcoming DIA-NN 2.8 Enterprise: - A major increase in identification numbers, in particular with InfinDIA: 25%+ on some datasets compared to 2.7. - Support for three additional technologies beyond ordinary DIA and DDA. Guess which ones? Back in the days of DIA-NN 1.8.1, released in early 2022, I genuinely thought we were close to the ceiling of how deeply we could see into the data. Turns out, that could not have been further from the truth. We have achieved massive gains in proteomic depth over the past year, and the pace is accelerating. Each advance is a step not only towards properly utilising the full potential of the very expensive mass spectrometry instrumentation, but also towards our ability to extract more knowledge from existing data. Automated reanalysis of ProteomeXchange repositories at scale - who knows what we might find there?

Means everyone can now get the full package for proteomics from Thermo: the MS + Evosep + DIA-NN!

Can AI take proteomics to a virtual cell? Our new perspective maps the path — from better peptide/protein ID to spatial & perturbation proteomics to AI Virtual Cells — and calls for an AI-friendly data ecosystem to get there. @guomics and Zhang lab. nature.com/articles/s41592-0…

Maths should match biology. Using ageing-brain redox dataset, this paper shows how tiny shifts in average oxidation can hide major proteome reorganisation. Even outside computational biology, its ideas may change how you think about (redox) biology. cell.com/patterns/fulltext/S…

And the preprint link is biorxiv.org/content/10.64898…

CysNet for oxiform inference from peptide level redox proteomics data is available now github.com/JamesCobley/CysNe… the preprint is to follow

We preprinted a method of measuring phosphosite stoichimetries (occupancies) in just two mass spectrometry acquisitions, using internal stable isotope labelled controls. In fact, we measure phospho stoichiometries in two ways, using dephosphorylation and using known relative ionisation efficiencies of phosphopeptides, compared to their unmodified forms: the measurements correlate well. In either case, estimates are well below on average compared to what was reported in some previous works. Link in comments below.

Our machine learning driven quantification algorithm for proteomics now published in Nature Biotechnology. Optimal protein quantification and, for the first time, error estimates for individual quantities. Paper link in comments below.

QuantUMS is the default method in our DIA-NN software. nature.com/articles/s41587-0…

Cleland immunoblotting can be used to quantify cysteine proteoform ensembles sciencedirect.com/science/ar…

We are delighted to present DIA-NN 2.5. Redefining LC-MS proteomics with up to 70% more proteins identified and powerful capabilities for statistics, visualisation and biological interpretation.

Nature Aging published a great piece presenting perspectives on the aging field from ~30 leading experts. Few areas of research make it as challenging to separate good science from hype as aging, which makes this article particularly valuable: nature.com/articles/s43587-0…

Biology isn’t just chemistry, it’s geometry in motion. This new Nature Biotech work shows that when we change the envelope geometry of a delivery particle, the efficiency of gene transfer into stem cells shifts dramatically. In VFD, this is expected: Geometry → Field Coherence → Cellular Response. Cells don’t react to molecules alone, they react to harmonic structure. When the geometry aligns, the information flows. VFD-BioSim models this directly, connecting φ-geometry, torsion, and receptor coherence into a unified biological framework. The closer biology moves toward geometry, the clearer the deeper mechanism becomes. @NatureBiotech @Nature #VFD #BioSim #GeometryOfLife #GeneEditing #Biotech @drmichaellevin @StuartHameroff

We acquired a large-scale mixed-species benchmark, with variable background, to comprehensively assess quantitative accuracy of proteomics. Key features: - 192 runs, 0.75ng - 15ng of variable human cell line backgroud. - Can dissect the impact of both random and systematic errors. - Can test how quantitation algorithms scale with experiment size and sample heterogeneity. Our insights based on the data: biorxiv.org/content/10.64898…. PRIDE repo will be made public in the next days.

Solving the computational challenge of phosphoproteomics with 𝐏𝐡𝐨-𝐓𝐢𝐩: dephosphorylation on-tip identifies the sequences of phosphorylated peptides. This serves as a basis for predicted spectral libraries, reducing the search space 10x-20x. biorxiv.org/content/10.1101/…

Is “oxidative stress” a useful concept in biology? In this work, led by Nikos, we argue that the term “oxidative stress” is devoid of biological meaning and, even worse, misleading.

If you believe that the current framework of redox biology (oxidative stress/redox signalling) is vague, then the approach introduced in this paper offers a geometric/physical basis for how redox modifications (cysteine oxidation) happen, interact, and shape biological outcomes.

Was a great pleasure to contribute to this work pubs.acs.org/doi/10.1021/jas…. DIA-NN release with support soon.