Skip to main content
Search

University of Turin

4 related articles

Subscribe to updates
  1. bioRxiv

    Protein stability predictors no longer contradict themselves: EvoPhi-DDG turns an alignment model into a potential, so forward and reverse scores cancel exactly

    Protein stability-change predictors often disagree with themselves: swapping the two sequences does not flip the sign of the prediction, and paths that detour through intermediate states at the same site fail to close. Without retraining, EvoPhi-DDG rewrites the family-alignment predictor MSAesm_ddG as a difference of an effective potential conditioned on the mutation site and alignment context, giving exact sign reversal and exact same-site additivity under a fixed alignment context. Averaged with equal weights with the published JanusDDG, it reaches a Pearson correlation of 0.59 on S669, against 0.55 for JanusDDG and 0.53 for EvoPhi-DDG.

  2. Nanoscale

    Hydrogen sulphide on refractory oxides: a foundation model screens configurations while DFT spectra reproduce the experimental band order

    Whether hydrogen sulphide physisorbs or dissociates on nanoscale refractory oxide surfaces had been hard to settle because the configuration space is too large to enumerate. This work pairs unbiased automated adsorption sampling with an off-the-shelf atomistic foundation model: MACE's MPA screens candidate minima, each survivor is re-optimised with PBE-D2*(N) and HSE06 DFT, and DFT multiconfigurational infrared spectra are compared with in situ spectra at 150 K. The computed spectra reproduce the red-shift order of the H–S stretch across the three materials: silica at 2603 cm⁻¹, forsterite at 2588 cm⁻¹ and MgO at 2504 cm⁻¹.

  3. arXiv

    Ranking positions by chat structure lets auditors explain 5% of tokens and keep nearly all threat-detection success

    Across about 4.7 million natural language autoencoder (NLA) explanations on four models and four auditing datasets, the study compares thirteen position-scoring signals computable in a single forward pass against a ranker trained only on chat structure, finding that chat structure beats the best individual signal in twelve of fourteen model-dataset cells, that explaining just 5% of positions retains nearly all of the success rate from explaining every position on three of four datasets, and that pretrained verbalizers recover words a fine-tuned model learned to conceal without additional verbalizer training.

  4. Urodynamic Report Writer (URapp): A GPT-4o-Based App for Urodynamic Interpretation and Draft Report Generation

    This study developed and validated the Urodynamic Report Writer App (URapp), a GPT-4o-based virtual assistant modeled on scientific literature and international urodynamic guidelines, fine-tuned under supervision by two expert urodynamicists using examinations from multiple referral centers, and evaluated by functional urology specialists on a 0-5 Likert scale, scoring high on interpretative accuracy (4.5 ± 0.2), explanation of urodynamic concepts (5.0 ± 0), adherence to evidence-based standards (4.8 ± 0.3), clarity and practicality of recommendations (4.2 ± 0.3), clinical relevance (3.8 ± 0.3), and diagnostic usefulness (3.5 ± 0.5), with all 100 consecutive urodynamic studies (median age 61 yr) receiving overall scores ≥4.

You’re all caught up