An official website of the United States government
Abstract With growing calls for increased surveillance of antibiotic resistance as an escalating global health threat, improved bioinformatic tools are needed for tracking antibiotic resistance genes (ARGs) across One Health domains. Most studies to date profile ARGs using sequence homology, but such approaches provide limited information about the broader context or function of the ARG in bacterial genomes. Here we introduce a new pipeline for identifying ARGs in genomic data that employs machine learning analysis of Protein-Protein Interaction Networks (PPINs) as a means to improve predictions of ARGs while also providing vital information about the context, such as gene mobility. A random forest model was trained to effectively differentiate between ARGs and nonARGs and was validated using the PPINs of ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, andEnterobacter cloacae), which represent urgent threats to human health because they tend to be multi-antibiotic resistant. The pipeline exhibited robustness in discriminating ARGs from nonARGs, achieving an average area under the precision-recall curve of 88%. We further identified that the neighbors of ARGs, i.e., genes connected to ARGs by only one edge, were disproportionately associated with mobile genetic elements, which is consistent with the understanding that ARGs tend to be mobile compared to randomly sampled genes in the PPINs. This pipeline showcases the utility of PPINs in discerning distinctive characteristics of ARGs within a broader genomic context and in differentiating ARGs from nonARGs through network-based attributes and interaction patterns. The code for running the pipeline is publicly available athttps://github.com/NazifaMoumi/PPI-ARG-ESKAPE
more »
« less
This content will become publicly available on July 11, 2026
resLens: genomic language models to enhance antibiotic resistance gene detection
Abstract The pace of antibiotic resistance necessitates advanced tools to detect and analyze antibiotic resistance genes (ARGs). We presentresLens, a family of genomic language models (gLM) leveraging latent genomic representations for ARG detection and analysis. Unlike alignment-based methods constrained by reference databases,resLensfine-tunes pre-trained gLMs on curated ARG datasets, achieving superior performance across several evaluation scenarios, including when ARGs exhibit dissimilar sequences and mechanisms to those in reference databases.
more »
« less
- Award ID(s):
- 2109688
- PAR ID:
- 10627422
- Publisher / Repository:
- bioRxiv
- Date Published:
- Format(s):
- Medium: X
- Institution:
- bioRxiv
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Antibiotic resistance (AR) presents a global health challenge, necessitating an improved understanding of the ecology, evolution, and dissemination of antibiotic resistance genes (ARGs). Several tools, databases, and algorithms are now available to facilitate the identification of ARGs in metagenomic sequencing data; however, direct annotation of short-read data provides limited contextual information. Knowledge of whether an ARG is carried in the chromosome or on a specific mobile genetic element (MGE) is critical to understanding mobility, persistence, and potential for co-selection. Here we developed ARGContextProfiler, a pipeline designed to extract and visualize ARG genomic contexts. By leveraging the assembly graph for genomic neighborhood extraction and validating contexts through read mapping, ARGContextProfiler minimizes chimeric errors that are a common artifact of assembly outputs. Testing on real, synthetic, and semi-synthetic data, including long-read sequencing data from environmental samples, demonstrated that ARGContextProfiler offers superior accuracy, precision, and sensitivity compared to conventional assembly-based methods. ARGContextProfiler thus provides a powerful tool for uncovering the genomic context of ARGs in metagenomic sequencing data, which can be of value to both fundamental and applied research aimed at understanding and stemming the spread of AR. The source code of ARGContextProfiler is publicly available atGitHub.more » « less
-
Elkins, Christopher A. (Ed.)ABSTRACT Low- and middle-income countries (LMICs) bear the largest mortality burden of antibiotic-resistant infections. Small-scale animal production and free-roaming domestic animals are common in many LMICs, yet data on zoonotic exchange of gut bacteria and antibiotic resistance genes (ARGs) in low-income communities are sparse. Differences between rural and urban communities with regard to population density, antibiotic use, and cohabitation with animals likely influence the frequency of transmission of gut bacterial communities and ARGs between humans and animals. Here, we determined the similarity in gut microbiomes, using 16S rRNA gene amplicon sequencing, and resistomes, using long-read metagenomics, between humans, chickens, and goats in a rural community compared to an urban community in Bangladesh. Gut microbiomes were more similar between humans and chickens in the rural (where cohabitation is more common) than the urban community, but there was no difference for humans and goats in the rural versus the urban community. Human and goat resistomes were more similar in the urban community, and ARG abundance was higher in urban animals than rural animals. We identified substantial overlap of ARG alleles in humans and animals in both settings. Humans and chickens had more overlapping ARG alleles than humans and goats. All fecal hosts from the urban community and rural humans carried ARGs on chromosomal contigs classified as potentially pathogenic bacteria, including Escherichia coli , Campylobacter jejuni , Clostridioides difficile , and Klebsiella pneumoniae . These findings provide insight into the breadth of ARGs circulating within human and animal populations in a rural compared to urban community in Bangladesh. IMPORTANCE While the development of antibiotic resistance in animal gut microbiomes and subsequent transmission to humans has been demonstrated in intensive farming environments and high-income countries, evidence of zoonotic exchange of antibiotic resistance in LMIC communities is lacking. This research provides genomic evidence of overlap of antibiotic resistance genes between humans and animals, especially in urban communities, and highlights chickens as important reservoirs of antibiotic resistance. Chicken and human gut microbiomes were more similar in rural Bangladesh, where cohabitation is more common. Incorporation of long-read metagenomics enabled characterization of bacterial hosts of resistance genes, which has not been possible in previous culture-independent studies using only short-read sequencing. These findings highlight the importance of developing strategies for combatting antibiotic resistance that account for chickens being reservoirs of ARGs in community environments, especially in urban areas.more » « less
-
null (Ed.)Wastewater treatment plants (WWTPs) receive a confluence of sewage containing antimicrobials, antibiotic resistant bacteria, antibiotic resistance genes (ARGs), and pathogens and thus are a key point of interest for antibiotic resistance surveillance. WWTP monitoring has the potential to inform with respect to the antibiotic resistance status of the community served as well as the potential for ARGs to escape treatment. However, there is lack of agreement regarding suitable sampling frequencies and monitoring targets to facilitate comparison within and among individual WWTPs. The objective of this study was to comprehensively evaluate patterns in metagenomic-derived indicators of antibiotic resistance through various stages of treatment at a conventional WWTP for the purpose of informing local monitoring approaches that are also informative for global comparison. Relative abundance of total ARGs decreased by ∼50% from the influent to the effluent, with each sampling location defined by a unique resistome (i.e., total ARG) composition. However, 90% of the ARGs found in the effluent were also detected in the influent, while the effluent ARG-pathogen taxonomic linkage patterns identified in assembled metagenomes were more similar to patterns in regional clinical surveillance data than the patterns identified in the influent. Analysis of core and discriminatory resistomes and general ARG trends across the eight sampling events (i.e., tendency to be removed, increase, decrease, or be found in the effluent only), along with quantification of ARGs of clinical concern, aided in identifying candidate ARGs for surveillance. Relative resistome risk characterization further provided a comprehensive metric for predicting the relative mobility of ARGs and likelihood of being carried in pathogens and can help to prioritize where to focus future monitoring and mitigation. Most antibiotics that were subject to regional resistance testing were also found in the WWTP, with the total antibiotic load decreasing by ∼40–50%, but no strong correlations were found between antibiotics and corresponding ARGs. Overall, this study provides insight into how metagenomic data can be collected and analyzed for surveillance of antibiotic resistance at WWTPs, suggesting that effluent is a beneficial monitoring point with relevance both to the local clinical condition and for assessing efficacy of wastewater treatment in reducing risk of disseminating antibiotic resistance.more » « less
-
null (Ed.)High-throughput sequencing is widely used for strain detection and characterization of antibiotic resistance in microbial metagenomic samples. Current analytical tools use curated antibiotic resistance gene (ARG) databases to classify individual sequencing reads or assembled contigs. However, identifying ARGs from raw read data can be time consuming (especially if assembly or alignment is required) and challenging, due to genome rearrangements and mutations. Here, we present the k-mer-based antibiotic gene resistance analyzer (KARGA), a multi-platform Java toolkit for identifying ARGs from metagenomic short read data. KARGA does not perform alignment; it uses an efficient double-lookup strategy, statistical filtering on false positives, and provides individual read classification as well as covering of the database resistome. On simulated data, KARGA’s antibiotic resistance class recall is 99.89% for error/mutation rates within 10%, and of 83.37% for error/mutation rates between 10% and 25%, while it is 99.92% on ARGs with rearrangements. On empirical data, KARGA provides higher hit score (≥1.5-fold) than AMRPlusPlus, DeepARG, and MetaMARC. KARGA has also faster runtimes than all other tools (2x faster than AMRPlusPlus, 7x than DeepARG, and over 100x than MetaMARC). KARGA is available under the MIT license at https://github.com/DataIntellSystLab/KARGA.more » « less
