Biology subjects

Hinic, V.

Publications and source records attributed to Hinic, V..

3 recordsLinked to original sources

bioRxiv2024

GPT-4 based AI agents - the new expert system for detection of antimicrobial resistance mechanisms?

BackgroundEUCAST recommends a two-step process for beta-lactamases in Gram-negative bacteria. Screening with minimal inhibitory concentrations (MICs) or inhibition zone diameters for potential extended spectrum beta-lactamase (ESBL), plasmid-mediated AmpC beta-lactamase, or carbapenemase production is followed by confirmatory tests. GPT-4 and its newly released customized GPT-agent may support the initial EUCAST-screening process. We aimed to validate a customized GPT-agent to identify potential resistance mechanisms. MethodsWe used 225 Gram-negative isolates. Based on phenotypic resistances against beta-lactam antibiotics, we formed four categories: "none", "ESBL", "AmpC", or "carbapenemase". We included 862 phenotypic categories. Next, we customized a GPT-agent with EUCAST-guidelines, expert rules, and EUCAST-breakpoint table (v13.1). We compared routine diagnostic outputs (reference) to (i) EUCAST-GPT-expert, (ii) medical microbiologists, and (iii) GPT-4 without customization. We determined performance as sensitivities and specificities to flag suspect resistance mechanisms. ResultsThree human readers showed concordance in 814/862 (94.4%) phenotypic categories and used in median eight words (IQR 4-11) for reasoning. Median sensitivity and specificity for ESBL, AmpC, and carbapenemase were 98%/99.1%, 96.8%/97.1%, and 95.5%/98.5%, respectively. Three independent prompting rounds of the GPT-agent showed concordance in 706/862 (81.9%) categories but used in median 158 words (IQR 140-174) for reasoning. Median sensitivity and specificity for ESBL, AmpC, and carbapenemase prediction were 95.4%/69.23%, 96.9%/86.3%, and 100%/98.8%, respectively. In the non-customized GPT-4, 169/862 (19.6%) categories could be interpreted. Of these 137/169 (81.1%) categories agreed with routine diagnostic. The non-customized GPT-4 used in median 85 words (IQR 72-105) for reasoning. ConclusionHuman experts showed higher concordance and shorter argumentations compared to GPT-agents. Human experts showed comparable median sensitivities and higher specificities compared to GPT-agents. GPT-agents showed more unspecific flagging of ESBL and AmpC, potentially, resulting in additional testing, diagnostic delays, and higher costs. GPT-4 and GPT-agents are not IVDR/FDA-approved, but validation of LLMs is critical and datasets for benchmarking are needed.

microbiology↗

bioRxiv2019

Impact of different amino acid substitutions in penicillin-binding protein 3 on beta-lactam susceptibility in Haemophilus influenzae

PurposeBeta-lactam antibiotics in combination with a beta-lactamase inhibitor are the first-line treatment option for Haemophilus influenzae infections. However, beta-lactamase-independent resistance to beta-lactams is increasing. This resistance mechanism has been linked to amino acid substitutions in the penicillin-binding protein 3 (PBP3), but how these substitutions lead to decreased binding affinities to certain beta-lactam antimicrobials remains unknown. MethodsWe investigated beta-lactam resistance and amino acid substitutions in PBP3 from fifty-three clinical isolates of H. influenzae collected in Switzerland from January to April 2016. Identification of key polymorphisms and classification of strains into PBP3 amino acid substitution groups I, II, and M was done as previously described. Based on published PBP3 crystal structures, we investigated how the group-specific amino acid substitutions impact the beta-lactam binding site. ResultsWe found that both group I and group II substitutions disrupt the Asn526-Arg517-Glu324 interaction, which might affect the configuration of the beta-lactam binding site. Amino acid substitutions in group M strains are distant from the active site and have most likely no impact on beta-lactam binding. In accordance with this observation, all group M strains showed minimal inhibitory concentrations (MICs) within the susceptible range for all tested antimicrobials and were not significantly different to the wild type (beta-lactamase producers excluded), while group I and group II strains showed significantly higher MICs for beta-lactam antimicrobials. ConclusionGroup M strains are phenotypically equal to the wild type, while amino acid substitutions of group I and group II might affect the beta-lactam binding through a common mechanism by disrupting the Asn526-Arg517-Glu324 interaction.

microbiology↗

bioRxiv2018

Resolving structural diversity of Carbapenemase-producing gram-negative bacteria using single molecule sequencing

Carbapenemase-producing bacteria are resistant against almost all commonly used betalactam and cephalosporin antibiotics and represent a growing public health crisis. Carbapenemases reside predominantly in mobile genetic elements and rapidly spread across genetic backgrounds and species boundaries. Here, we report more than one hundred finished, high quality genomes of carbapenemase producing enterobacteriaceae, P. aeruginosa and A. baumannii sequenced with Oxford Nanopore and Illumina technologies. We developed a number of high-throughput criteria to assess the quality of fully assembled genomes for which curated references do not exist. Using this diverse collection of closed genomes and plasmids, we demonstrate rapid movement of carbapenemase between genomic neighborhoods, sequence types, and across species boundaries with distinct patterns for different carbapenemases. Lastly, we present evidence of multiple ancestral recombination events between different Enterobacteriaceae MLSTs. Taken together, our samples suggest a hierarchical picture of genomic variation produced by the evolution of carbapenemase producing bacteria that will require new models to adequately understand and track.

microbiology↗