Progress

The AI Illusion

admin

Why Grassroots Research Demands Open-Source Accountability. Generative AI promises unprecedented capabilities in data analysis, but it introduces severe risks for academic research. While large language models can accelerate evidence synthesis, they frequently fabricate plausible-sounding but non-existent citations. The statistical reality is alarming. A recent comparative analysis revealed that when conducting systematic reviews, GPT-3.5 hallucinated 39.6% of its references, GPT-4 hallucinated 28.6%, and Bard reached a staggering 91.4%.

Furthermore, these generative AI tools miss a median of 91% of relevant studies compared to human researchers, making incorrect inclusion decisions in up to 29% of instances and data extraction errors in up to 31% of cases. Because of these exceptionally high error rates, constant human oversight is strictly mandatory. To combat unverified AI-driven workflows, the scientific community requires robust infrastructure. Look out for ResinTox Tools: an upcoming platform specifically dedicated to democratizing grassroots research. By hosting in-house and open source code, applications, and other tailored resources, ResinTox Tools aims to empower citizen scientists to use AI safely. Ultimately, promoting open-source accountability is essential. Countering AI hallucinations will require users of community-driven platforms to ensure that our future research methodologies remain safe, responsible, and rigorously verified.

Chelli, M., Descamps, J., Lavoué, V., Trojani, C., Azar, M., Deckert, M., Raynier, J. L., Clowez, G., Boileau, P., & Ruetsch-Chelli, C. ‘Hallucination Rates and Reference Accuracy of ChatGPT and Bard for Systematic Reviews: Comparative Analysis’
Source Access

Clark, J. et al. ‘Generative artificial intelligence use in evidence synthesis: A systematic review’
Source Access

AlphaFold and Machine Learning

admin

A Paradigm Shift in Environmental Toxicology. Environmental toxicology is undergoing a paradigm shift. For years, researchers relied on traditional homology modeling tools like SWISS-MODEL to predict protein structures. Today, the integration of AlphaFold’s highly accurate ab initio 3D predictions is transforming how we assess environmental risks. A prime example is the screening of endocrine-disrupting chemicals (EDCs). Recent analyses utilizing the zebrafish androgen receptor demonstrate that AlphaFold-generated models offer superior structural stability.

Moreover, AlphaFold-derived MMPBSA binding energies are statistically significant (p < 0.05) contributors to ligand-specific variance when compared to in vitro EC50 values, easily outperforming older techniques. Beyond direct screening, toxicologists are leveraging the expansive AlphaFold-2 database, which houses over 350,000 predicted protein structures, to train novel machine learning pipelines. These computational frameworks accurately quantify 3D protein similarities across diverse species. By mapping cross-species structural alignments, researchers can now optimize the selection of animal models for human toxicity testing with unprecedented precision. Ultimately, the fusion of high-fidelity structural predictions and advanced machine learning is elevating the accuracy of environmental risk assessments, moving the field past traditional limitations into a highly predictive, data-driven future.

Md Adnan Karim, Chang Gyun Park, Hyunki Cho, Annmariya Elayanithottathil Sebastian, Chang Seon Ryu, Juyong Yoon, Young Jun Kim ‘Leveraging AlphaFold models to predict androgenic effects of endocrine-disrupting chemicals through zebrafish androgen receptor analysis’
Source Access

Shreyas U Hirway, Xiao Xu, Fan Fan ‘A novel computational machine learning pipeline to quantify similarities in 3D protein structures’
Source Access

Beyond the Benchmark

admin

How Organoids are Revolutionizing Environmental Toxicology
Modern environmental contaminants pose an unprecedented threat to human health. Alarmingly, individuals consume up to 52,000 microplastic particles annually through diet alone, allowing these pollutants to penetrate systemic circulation, accumulate in vital organs like the lungs and placenta, and drive chronic inflammation. Exacerbating this crisis is the reality of environmental exposures: we encounter complex, synergistic chemical mixtures rather than isolated toxins. Traditional single-chemical regulatory frameworks struggle to keep pace, as seen in remediation efforts where activated carbon removes up to 98% of long-chain PFAS but captures less than 60% of short-chain variants.

To properly address these intertwined threats, toxicology is undergoing a paradigm shift. Advanced 3D human organoids and organ-on-a-chip technologies are bypassing the translational and ethical limitations of traditional animal testing by offering highly accurate, physiologically relevant platforms. For example, recent assessments utilizing human cortical organoids successfully revealed how a defined mixture of 21 endocrine-disrupting chemicals severely impairs neuronal migration and brain development. As chemical exposures grow increasingly complex, integrating these revolutionary, human-derived models into mainstream toxicology is no longer optional. It is an urgent necessity to accurately evaluate synergistic toxicities and safeguard global public health.

References:
Chengyu Hu, Sheng Yang, Tianyi Zhang ‘Organoids and organoids-on-a-chip as the new testing strategies for environmental toxicology-applications & advantages’ Source Access

Xu Zhang, Chunhong Yu, Peng Wang, Chunping Yang ‘Microplastics and human health: unraveling the toxicological pathways and implications for public health’ Source Access

ResinTox Fallout Predictor

admin

The Fallout Predictor is a pioneering web-based platform designed to model the dispersion of hazardous substances with unprecedented precision. Moving beyond the limitations of legacy software, this application utilises a sophisticated Gaussian-Lagrangian hybrid approach to simulate plumes interacting with complex geographic and urban landscapes. Large volume users are advised to become subscribers to the Open Meteo API-service, as the system will have limited amount of calls per month. With their api key, it is possible to make more predictions (or more playing). For every day usage or testing purposes however, the free tier is usually enough. Chemical data is fetched from PubChem REST API by default, but with your Google Gemini API-Key, it is possible to fetch data for chemicals, compounds and other substances too. Secure API-Key management is handled internally on the client side (Browser).

One of the most impressive features is the real-time integration of OpenStreetMap data, allowing for plume deflection and recursive branching when encountering architectural mass. This provides responders with a far more realistic visualisation than traditional linear projections. Furthermore, the platform is currently in a Limited Time Open Access Beta, offering an interactive Physics Sandbox for users to explore the underlying engine. Do not hesitate to provide feedback, so we can release a full feature application before the end of 2026.

The future looks even brighter with the upcoming implementation of Gemini 3.0 (With API-Key) logic. This AI enhancement will automate chemical property retrieval and provide rapid geospatial population estimates, significantly reducing critical decision-making time during emergencies. Whilst current standards like ALOHA offer basic insights, the Fallout Predictor v2.5.0 sets a new benchmark in emergency readiness and situational awareness. For organisations looking to bolster their safety protocols, this programme represents a vital leap forward in predictive technology.

Advancing Toxicology:

admin

The ResinTox Mission

At ResinTox, our mission is to empower the global toxicology community through cutting-edge software solutions, robust data, and rigorous evidence-based research. Guided by the principle of being ‘By People – For People’, we strive to bridge the gap between complex scientific data and the practical needs of healthcare providers, regulatory agencies, and corporations.

Pioneering Progress: The RAPID System

Our most significant breakthrough is the RAPID system (Rapid On-site Analyte-specific Peptide Intervention and Diversion), designed to revolutionise snakebite envenomation management. Traditionally, snakebite treatment has been hindered by diagnostic uncertainty and cold-chain logistics. Our four-pillar approach—combining on-site portable diagnostics, cloud-based AI analysis, and on-demand peptide synthesis—aims to provide personalised, life-saving care even in the most remote locations.

Tools for the Future

Beyond snakebite management, we are making strides in in-silico toxicology. The development of the ResinTox Toxicology Toolkit provides an all-in-one platform for symptom prediction and de novo molecule development. Whether we are conducting meta-analyses on the health effects of nicotine or investigating venom variability in Vipera berus, our commitment remains the same: ensuring that the highest standard of scientific evidence drives global decisions on chemical safety and public health.

The journey toward a safer, more informed world is well underway.

An Adder’s Bite, A Peptide’s Fight: A Framework for Peptide-Based Neutralisation of Vipera berus Phospholipase A₂

admin

There is a new Preprint available that is now undergoing peer review. You can read the draft and manuscript here: https://doi.org/10.5281/zenodo.17514183

Abstract

Across Eurasia, envenomation by the common European adder, Vipera berus, presents a significant medical challenge, primarily driven by phospholipase A₂ (PLA₂) enzymes. Traditional antivenom therapies are hampered by specificity, immunogenicity, and logistical limitations. This article, structured into distinct chapters, first establishes the biochemical challenge posed by Vipera berus venom and the central role of PLA₂. It then outlines a focused, structure-based strategy for designing synthetic peptide inhibitors that specifically target the primary toxic PLA₂ isoform in V. berus venom (UniProt: P31854). Drawing from successful precedents in inhibiting homologous viper and elapid PLA₂s, we propose a design blueprint for short peptides (5-7 amino acids) that function as high-affinity active site blockers. The proposed mechanism involves a hydrophobic peptide scaffold that anchors the inhibitor within the enzyme’s substrate-binding channel, combined with a strategically placed “warhead” residue (e.g., Tyrosine or Arginine) designed to form high-energy hydrogen bonds or salt bridges with the catalytic dyad (His48/Asp49). This approach effectively neutralises the enzyme’s toxic activity. The clinical translation of such designer peptides is made feasible by emerging on-demand manufacturing platforms, such as the RAPID system, which utilise rapid synthesis technologies to produce patient-specific therapeutics. This peptide-centric framework offers a pathway toward a new generation of precise, safe, and logistically robust antivenoms.