lusi

[Biology+AI Daily @BiologyAIDaily]

Explainable AI for end-to-end pathogen target discovery and molecular design

1 APEX introduces a unified explainable AI framework combining ESM-2 protein embeddings, graph attention networks, and diffusion models to bridge target identification and inhibitor design in a single pipeline.

2 The dual-model architecture features APEX-Tar for pathogen-specific essentiality/virulence prediction and APEX-Drug as a universal druggability classifier, enabling cross-species proteome-scale target discovery.

3 Built-in interpretability through attention weights and GNNExplainer provides residue-level mechanistic insights, directly guiding structure-based diffusion models to generate pocket-specific inhibitors without black-box limitations.

4 Applied to Botrytis cinerea, APEX recovered known fungal targets including endopolygalacturonase 1 and Hog1 MAPK, while identifying 805 high-confidence novel candidates enriched in necrotrophic virulence functions.

5 The pipeline successfully designed candidate inhibitors for GmrSD, a top-ranked fungal target, with predicted binding energies of -7.6 to -7.3 kcal/mol and favorable drug-like properties.

6 Cross-kingdom generalization was demonstrated by retraining APEX-Tar on bacterial virulence factors and applying it to Acinetobacter baumannii, where YadV fimbrial chaperone emerged as the top target with a previously undescribed druggable pocket.

7 The YadV case revealed both the canonical pilicide-binding groove and a novel pocket distinct from known sites, with generated inhibitors showing strong predicted affinities (-8.9 and -8.8 kcal/mol) and distinct interaction patterns.

8 This work addresses the critical bottleneck in antimicrobial discovery by integrating interpretable target prioritization with rational molecular generation, offering a scalable blueprint for addressing resistance challenges in agriculture and medicine.

💻Code: https://t.co/f9XQji9aBL
📜Paper: https://t.co/h9hHJX9Up6
#ExplainableAI #DrugDiscovery #AntimicrobialResistance #GraphNeuralNetworks #ProteinLanguageModels #MolecularDesign #Bioinformatics #ComputationalBiology

[FerrandizLab @FerrandizLab]

Ofrecemos tres contratos de 2 años del #ProgramaInvestigo2023 de la @GVA_innova . ¿Te interesa la Biotecnología de Plantas 🌿🌸🍅y cumples los requisitos 👇👇? ¡Ven a trabajar con nosotros al @IBMCP!

[ÚRSULA @ursuulasanchez]

Hola! Somos dos chicas que buscamos compañer@ de piso en Sevilla para el curso 2023/24. Zona La Macarena, cerca de facultades, bien comunicado. 217 con comunidad y agua. Se agradece difusión.

[lusi @lusijim]

como smp en mi vida, palante con to

[lusi @lusijim]

es el bru

[no me llamo vicenta @kansaita_]

alguien: no entiendo por qué no te gusta tomar el sol

yo cuando me da el sol:

[lusi @lusijim]

24 años intentando q me gusten los garbanzos no puedo más

[lusi @lusijim]

adrenalina dame paz

[lusi @lusijim]

@pxblomarti pero más solete

[Ainhoa @lasinverwenza]

yo toda la vida pensando que los adultos compraban lotería de navidad por la ilusión y resulta que la compran porque su compi de trabajo tiene el mismo número y cómo le toque al hijo de puta ese y yo tenga que seguir currando me mato, vaya

[lusi @lusijim]

flepa

[lusi @lusijim]

https://t.co/RzgxGKGfWk

[lusi @lusijim]

tengo un serio problema con maría escoté yo e

[lusi @lusijim]

madre mía lendakaris en 5 días selia

[lusi @lusijim]

libres son los pensamientos

[✰ @xvbrenda]

i get the ick so fast, it’s crzy

[lusi @lusijim]

@inespuli why tf would you do that

[lusi @lusijim]

odio el azul marino no lo soporto no sé explicarlo