Machine learning based on pangenome-wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogens
Abstract Variations in host species significantly impact bacterial growth traits and antibiotic resistance, making it essential to consider host origin when evaluating the zoonotic potential of pathogens. This study focuses on multiple Brucella species, which share highly similar genetic material, t...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-08-01
|
| Series: | Communications Biology |
| Online Access: | https://doi.org/10.1038/s42003-025-08650-3 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849226028180832256 |
|---|---|
| author | Cheng Han Shiying Lu Pan Hu Jiang Chang Deying Zou Feng Li Yansong Li Qiang Lu Honglin Ren |
| author_facet | Cheng Han Shiying Lu Pan Hu Jiang Chang Deying Zou Feng Li Yansong Li Qiang Lu Honglin Ren |
| author_sort | Cheng Han |
| collection | DOAJ |
| description | Abstract Variations in host species significantly impact bacterial growth traits and antibiotic resistance, making it essential to consider host origin when evaluating the zoonotic potential of pathogens. This study focuses on multiple Brucella species, which share highly similar genetic material, to explore the relationship between host origin and zoonotic potential by integrating pan-genome-wide association studies (pan-GWAS) with machine learning (ML). Our results present an open pangenome of Brucella spp. derived from the whole-genome sequencing (WGS) data of 991 strains and identify 268 genes potentially associated with the zoonotic potential of Brucella. Integrating these genes into an ML model based on the support vector machine (SVM) algorithm allows us to predict the zoonotic potential of various Brucella strains with high accuracy. Our findings reveal that zoonotic potential varies by host origin: Brucella melitensis strains isolated from humans exhibit higher zoonotic potential than those isolated from cattle, goats, and sheep, while Brucella suis biovar 2 strains isolated from domestic pigs display higher zoonotic potential than those isolated from wild boars. Our study proposes a method for predicting and quantifying the zoonotic potential of closely related bacterial pathogens from different host origins, providing valuable insights for risk assessment and public health strategy. |
| format | Article |
| id | doaj-art-ec718f2260d7449b92c31460cdf0b682 |
| institution | Kabale University |
| issn | 2399-3642 |
| language | English |
| publishDate | 2025-08-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Communications Biology |
| spelling | doaj-art-ec718f2260d7449b92c31460cdf0b6822025-08-24T11:46:18ZengNature PortfolioCommunications Biology2399-36422025-08-018111310.1038/s42003-025-08650-3Machine learning based on pangenome-wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogensCheng Han0Shiying Lu1Pan Hu2Jiang Chang3Deying Zou4Feng Li5Yansong Li6Qiang Lu7Honglin Ren8State Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityState Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityState Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityState Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityState Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityState Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityState Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityState Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityState Key Laboratory for Diagnosis and Treatment of Severe Zoonotic Infectious Diseases, Key Laboratory for Zoonosis Research of the Ministry of Education, Institute of Zoonosis, College of Veterinary Medicine, Chongqing Research Institute, Jilin UniversityAbstract Variations in host species significantly impact bacterial growth traits and antibiotic resistance, making it essential to consider host origin when evaluating the zoonotic potential of pathogens. This study focuses on multiple Brucella species, which share highly similar genetic material, to explore the relationship between host origin and zoonotic potential by integrating pan-genome-wide association studies (pan-GWAS) with machine learning (ML). Our results present an open pangenome of Brucella spp. derived from the whole-genome sequencing (WGS) data of 991 strains and identify 268 genes potentially associated with the zoonotic potential of Brucella. Integrating these genes into an ML model based on the support vector machine (SVM) algorithm allows us to predict the zoonotic potential of various Brucella strains with high accuracy. Our findings reveal that zoonotic potential varies by host origin: Brucella melitensis strains isolated from humans exhibit higher zoonotic potential than those isolated from cattle, goats, and sheep, while Brucella suis biovar 2 strains isolated from domestic pigs display higher zoonotic potential than those isolated from wild boars. Our study proposes a method for predicting and quantifying the zoonotic potential of closely related bacterial pathogens from different host origins, providing valuable insights for risk assessment and public health strategy.https://doi.org/10.1038/s42003-025-08650-3 |
| spellingShingle | Cheng Han Shiying Lu Pan Hu Jiang Chang Deying Zou Feng Li Yansong Li Qiang Lu Honglin Ren Machine learning based on pangenome-wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogens Communications Biology |
| title | Machine learning based on pangenome-wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogens |
| title_full | Machine learning based on pangenome-wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogens |
| title_fullStr | Machine learning based on pangenome-wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogens |
| title_full_unstemmed | Machine learning based on pangenome-wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogens |
| title_short | Machine learning based on pangenome-wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogens |
| title_sort | machine learning based on pangenome wide association studies reveals the impact of host source on the zoonotic potential of closely related bacterial pathogens |
| url | https://doi.org/10.1038/s42003-025-08650-3 |
| work_keys_str_mv | AT chenghan machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens AT shiyinglu machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens AT panhu machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens AT jiangchang machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens AT deyingzou machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens AT fengli machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens AT yansongli machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens AT qianglu machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens AT honglinren machinelearningbasedonpangenomewideassociationstudiesrevealstheimpactofhostsourceonthezoonoticpotentialofcloselyrelatedbacterialpathogens |