Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics
Predicting the dynamics and functions of microbiomes constructed from the bottom-up is a key challenge in exploiting them to our benefit. Current models based on ecological theory fail to capture complex community behaviors due to higher order interactions, do not scale well with increasing complexi...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
eLife Sciences Publications Ltd
2022-06-01
|
| Series: | eLife |
| Subjects: | |
| Online Access: | https://elifesciences.org/articles/73870 |
| _version_ | 1828108752054648832 |
|---|---|
| author | Mayank Baranwal Ryan L Clark Jaron Thompson Zeyu Sun Alfred O Hero Ophelia S Venturelli |
| author_facet | Mayank Baranwal Ryan L Clark Jaron Thompson Zeyu Sun Alfred O Hero Ophelia S Venturelli |
| author_sort | Mayank Baranwal |
| collection | DOAJ |
| description | Predicting the dynamics and functions of microbiomes constructed from the bottom-up is a key challenge in exploiting them to our benefit. Current models based on ecological theory fail to capture complex community behaviors due to higher order interactions, do not scale well with increasing complexity and in considering multiple functions. We develop and apply a long short-term memory (LSTM) framework to advance our understanding of community assembly and health-relevant metabolite production using a synthetic human gut community. A mainstay of recurrent neural networks, the LSTM learns a high dimensional data-driven non-linear dynamical system model. We show that the LSTM model can outperform the widely used generalized Lotka-Volterra model based on ecological theory. We build methods to decipher microbe-microbe and microbe-metabolite interactions from an otherwise black-box model. These methods highlight that Actinobacteria, Firmicutes and Proteobacteria are significant drivers of metabolite production whereas Bacteroides shape community dynamics. We use the LSTM model to navigate a large multidimensional functional landscape to design communities with unique health-relevant metabolite profiles and temporal behaviors. In sum, the accuracy of the LSTM model can be exploited for experimental planning and to guide the design of synthetic microbiomes with target dynamic functions. |
| first_indexed | 2024-04-11T10:51:43Z |
| format | Article |
| id | doaj.art-25fdde5e7476427aa4756f294fab0c38 |
| institution | Directory Open Access Journal |
| issn | 2050-084X |
| language | English |
| last_indexed | 2024-04-11T10:51:43Z |
| publishDate | 2022-06-01 |
| publisher | eLife Sciences Publications Ltd |
| record_format | Article |
| series | eLife |
| spelling | doaj.art-25fdde5e7476427aa4756f294fab0c382022-12-22T04:28:53ZengeLife Sciences Publications LtdeLife2050-084X2022-06-011110.7554/eLife.73870Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamicsMayank Baranwal0https://orcid.org/0000-0001-9354-2826Ryan L Clark1https://orcid.org/0000-0001-7865-2496Jaron Thompson2https://orcid.org/0000-0001-5967-0234Zeyu Sun3Alfred O Hero4https://orcid.org/0000-0002-2531-9670Ophelia S Venturelli5https://orcid.org/0000-0003-2200-1963Department of Systems and Control Engineering, Indian Institute of Technology, Bombay, India; Division of Data & Decision Sciences, Tata Consultancy Services Research, Mumbai, IndiaDepartment of Biochemistry, University of Wisconsin-Madison, Madison, United StatesDepartment of Chemical & Biological Engineering, University of Wisconsin-Madison, Madison, United StatesDepartment of Electrical Engineering & Computer Science, University of Michigan, Ann Arbor, United StatesDepartment of Electrical Engineering & Computer Science, University of Michigan, Ann Arbor, United States; Department of Biomedical Engineering, University of Michigan, Ann Arbor, United States; Department of Statistics, University of Michigan, Ann Arbor, United StatesDepartment of Biochemistry, University of Wisconsin-Madison, Madison, United States; Department of Chemical & Biological Engineering, University of Wisconsin-Madison, Madison, United States; Department of Bacteriology, University of Wisconsin-Madison, Madison, United StatesPredicting the dynamics and functions of microbiomes constructed from the bottom-up is a key challenge in exploiting them to our benefit. Current models based on ecological theory fail to capture complex community behaviors due to higher order interactions, do not scale well with increasing complexity and in considering multiple functions. We develop and apply a long short-term memory (LSTM) framework to advance our understanding of community assembly and health-relevant metabolite production using a synthetic human gut community. A mainstay of recurrent neural networks, the LSTM learns a high dimensional data-driven non-linear dynamical system model. We show that the LSTM model can outperform the widely used generalized Lotka-Volterra model based on ecological theory. We build methods to decipher microbe-microbe and microbe-metabolite interactions from an otherwise black-box model. These methods highlight that Actinobacteria, Firmicutes and Proteobacteria are significant drivers of metabolite production whereas Bacteroides shape community dynamics. We use the LSTM model to navigate a large multidimensional functional landscape to design communities with unique health-relevant metabolite profiles and temporal behaviors. In sum, the accuracy of the LSTM model can be exploited for experimental planning and to guide the design of synthetic microbiomes with target dynamic functions.https://elifesciences.org/articles/73870human gut microbiomeecological networkdynamical systemsmicrobiome engineeringmachine learningmicrobial metabolism |
| spellingShingle | Mayank Baranwal Ryan L Clark Jaron Thompson Zeyu Sun Alfred O Hero Ophelia S Venturelli Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics eLife human gut microbiome ecological network dynamical systems microbiome engineering machine learning microbial metabolism |
| title | Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics |
| title_full | Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics |
| title_fullStr | Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics |
| title_full_unstemmed | Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics |
| title_short | Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics |
| title_sort | recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics |
| topic | human gut microbiome ecological network dynamical systems microbiome engineering machine learning microbial metabolism |
| url | https://elifesciences.org/articles/73870 |
| work_keys_str_mv | AT mayankbaranwal recurrentneuralnetworksenabledesignofmultifunctionalsynthetichumangutmicrobiomedynamics AT ryanlclark recurrentneuralnetworksenabledesignofmultifunctionalsynthetichumangutmicrobiomedynamics AT jaronthompson recurrentneuralnetworksenabledesignofmultifunctionalsynthetichumangutmicrobiomedynamics AT zeyusun recurrentneuralnetworksenabledesignofmultifunctionalsynthetichumangutmicrobiomedynamics AT alfredohero recurrentneuralnetworksenabledesignofmultifunctionalsynthetichumangutmicrobiomedynamics AT opheliasventurelli recurrentneuralnetworksenabledesignofmultifunctionalsynthetichumangutmicrobiomedynamics |