1. Computational and Systems Biology
  2. Neuroscience

Training deep neural density estimators to identify mechanistic models of neural dynamics

  1. Pedro J Gonçalves  Is a corresponding author
  2. Jan-Matthis Lueckmann  Is a corresponding author
  3. Michael Deistler  Is a corresponding author
  4. Marcel Nonnenmacher
  5. Kaan Öcal
  6. Giacomo Bassetto
  7. Chaitanya Chintaluri
  8. William F Podlaski
  9. Sara A Haddad
  10. Tim Vogels
  11. David S Greenberg
  12. Jakob H Macke  Is a corresponding author
  1. Center of Advanced European Studies and Research (caesar), Germany
  2. Technical University of Munich, Germany
  3. Mathematical Institute, Germany
  4. University of Oxford, United Kingdom
  5. Max-Planck Institute for Brain Research, Germany
  6. University of Tübingen, Germany
https://doi.org/10.7554/eLife.56261
Share this article
Cite this article
  1. Pedro J Gonçalves
  2. Jan-Matthis Lueckmann
  3. Michael Deistler
  4. Marcel Nonnenmacher
  5. Kaan Öcal
  6. Giacomo Bassetto
  7. Chaitanya Chintaluri
  8. William F Podlaski
  9. Sara A Haddad
  10. Tim Vogels
  11. David S Greenberg
  12. Jakob H Macke
(2020)
Training deep neural density estimators to identify mechanistic models of neural dynamics
eLife 9:e56261.
https://doi.org/10.7554/eLife.56261
  2. Download BibTeX
  3. Download .RIS

Abstract

Mechanistic modeling in neuroscience aims to explain observed phenomena in terms of underlying causes. However, determining which model parameters agree with complex and stochastic neural data presents a significant challenge. We address this challenge with a machine learning tool which uses deep neural density estimators- trained using model simulations- to carry out Bayesian inference and retrieve the full space of parameters compatible with raw data or selected data features. Our method is scalable in parameters and data features, and can rapidly analyze new data after initial training. We demonstrate the power and flexibility of our approach on receptive fields, ion channels, and Hodgkin-Huxley models. We also characterize the space of circuit configurations giving rise to rhythmic activity in the crustacean stomatogastric ganglion, and use these results to derive hypotheses for underlying compensation mechanisms. Our approach will help close the gap between data-driven and theory-driven models of neural dynamics.

Data availability

The contributions of the work are primarily the development and application of computational models, no new data has been obtained or is being published. All code and associated data will be shared in the GitHub repository accompanying the article.

The following previously published data sets were used

Article and author information

Author details

  1. Pedro J Gonçalves

    Max Planck Research Group Neural Systems Analysis, Center of Advanced European Studies and Research (caesar), Bonn, Germany
    For correspondence
    pedro.goncalves@caesar.de
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6987-4836

  2. Jan-Matthis Lueckmann

    Department of Electrical and Computer Engineering, Technical University of Munich, Munich, Germany
    For correspondence
    jan-matthis.lueckmann@tum.de
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4320-4663

  3. Michael Deistler

    Department of Electrical and Computer Engineering, Technical University of Munich, Munich, Germany
    For correspondence
    michael.deistler@tum.de
    Competing interests
    The authors declare that no competing interests exist.

  4. Marcel Nonnenmacher

    Department of Electrical and Computer Engineering, Technical University of Munich, Munich, Germany
    Competing interests
    The authors declare that no competing interests exist.

  5. Kaan Öcal

    University of Bonn, Mathematical Institute, Bonn, Germany
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8528-6858

  6. Giacomo Bassetto

    Max Planck Research Group Neural Systems Analysis, Center of Advanced European Studies and Research (caesar), Bonn, Germany
    Competing interests
    The authors declare that no competing interests exist.

  7. Chaitanya Chintaluri

    Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, United Kingdom
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4252-1608

  8. William F Podlaski

    Physiology Anatomy and Genetics, University of Oxford, Oxford, United Kingdom
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6619-7502

  9. Sara A Haddad

    Neural Systems and Coding, Max-Planck Institute for Brain Research, Frankfurt, Germany
    Competing interests
    The authors declare that no competing interests exist.

  10. Tim Vogels

    Neural Systems and Coding, Max-Planck Institute for Brain Research, Frankfurt, Germany
    Competing interests
    The authors declare that no competing interests exist.

  11. David S Greenberg

    Department of Electrical and Computer Engineering, Technical University of Munich, Munich, Germany
    Competing interests
    The authors declare that no competing interests exist.

  12. Jakob H Macke

    Excellence Cluster Machine Learning, University of Tübingen, Tübingen, Germany
    For correspondence
    Jakob.Macke@gmail.com
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5154-8912

Funding

Deutsche Forschungsgemeinschaft (SFB~1233)

  • Jan-Matthis Lueckmann
  • Marcel Nonnenmacher
  • Giacomo Bassetto
  • Jakob H Macke

Deutsche Forschungsgemeinschaft (SFB~1089)

  • Pedro J Gonçalves
  • Jakob H Macke

Deutsche Forschungsgemeinschaft (SPP~2041)

  • Jakob H Macke

Bundesministerium für Bildung und Forschung (01IS18052 A-D)

  • Michael Deistler
  • Marcel Nonnenmacher
  • Jakob H Macke

Wellcome Trust and the Royal Society (WT100000)

  • William F Podlaski
  • Tim Vogels

H2020 European Research Council (SYNAPSEEK)

  • Chaitanya Chintaluri
  • William F Podlaski
  • Tim Vogels

Wellcome Trust Senior Research Fellowship (214316/Z/18/Z)

  • Tim Vogels

UK Research and Innovation, Biotechnology and Biological Sciences Research Council (UKRI-BBSRC BB/N019512/1)

  • Chaitanya Chintaluri

Deutsche Forschungsgemeinschaft (Germany's Excellence Strategy - EXC-Number 2064/1 610 - Project number 390727645)

  • Jakob H Macke

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Reviewing Editor

  1. Timothy O'Leary, University of Cambridge, United Kingdom

Version history

  1. Received: February 21, 2020
  2. Accepted: September 16, 2020
  3. Accepted Manuscript published: September 17, 2020 (version 1)
  4. Accepted Manuscript updated: September 18, 2020 (version 2)
  5. Version of Record published: October 22, 2020 (version 3)

Copyright

© 2020, Gonçalves et al.

This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 8,480
    views
  • 1,211
    downloads
  • 108
    citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Pedro J Gonçalves
  2. Jan-Matthis Lueckmann
  3. Michael Deistler
  4. Marcel Nonnenmacher
  5. Kaan Öcal
  6. Giacomo Bassetto
  7. Chaitanya Chintaluri
  8. William F Podlaski
  9. Sara A Haddad
  10. Tim Vogels
  11. David S Greenberg
  12. Jakob H Macke
(2020)
Training deep neural density estimators to identify mechanistic models of neural dynamics
eLife 9:e56261.
https://doi.org/10.7554/eLife.56261

Share this article

https://doi.org/10.7554/eLife.56261

Categories and tags

Research organisms

Further reading

    1. Computational and Systems Biology
    2. Neuroscience

    Interrogating theoretical models of neural computation with emergent property inference

    Sean R Bittner, Agostina Palmigiano ... John Cunningham
    Research Article

    A cornerstone of theoretical neuroscience is the circuit model: a system of equations that captures a hypothesized neural mechanism. Such models are valuable when they give rise to an experimentally observed phenomenon -- whether behavioral or a pattern of neural activity -- and thus can offer insights into neural computation. The operation of these circuits, like all models, critically depends on the choice of model parameters. A key step is then to identify the model parameters consistent with observed phenomena: to solve the inverse problem. In this work, we present a novel technique, emergent property inference (EPI), that brings the modern probabilistic modeling toolkit to theoretical neuroscience. When theorizing circuit models, theoreticians predominantly focus on reproducing computational properties rather than a particular dataset. Our method uses deep neural networks to learn parameter distributions with these computational properties. This methodology is introduced through a motivational example of parameter inference in the stomatogastric ganglion. EPI is then shown to allow precise control over the behavior of inferred parameters and to scale in parameter dimension better than alternative techniques. In the remainder of this work, we present novel theoretical findings in models of primary visual cortex and superior colliculus, which were gained through the examination of complex parametric structure captured by EPI. Beyond its scientific contribution, this work illustrates the variety of analyses possible once deep learning is harnessed towards solving theoretical inverse problems.

    1. Microbiology and Infectious Disease

    Interplay between bacterial deubiquitinase and ubiquitin E3 ligase regulates ubiquitin dynamics on Legionella phagosomes

    Shuxin Liu, Jiwei Luo ... Zhao-Qing Luo
    Research Article Updated

    Legionella pneumophila extensively modulates the host ubiquitin network to create the Legionella-containing vacuole (LCV) for its replication. Many of its virulence factors function as ubiquitin ligases or deubiquitinases (DUBs). Here, we identify Lem27 as a DUB that displays a preference for diubiquitin formed by K6, K11, or K48. Lem27 is associated with the LCV where it regulates Rab10 ubiquitination in concert with SidC and SdcA, two bacterial E3 ubiquitin ligases. Structural analysis of the complex formed by an active fragment of Lem27 and the substrate-based suicide inhibitor ubiquitin-propargylamide (PA) reveals that it harbors a fold resembling those in the OTU1 DUB subfamily with a Cys-His catalytic dyad and that it recognizes ubiquitin via extensive hydrogen bonding at six contact sites. Our results establish Lem27 as a DUB that functions to regulate protein ubiquitination on L. pneumophila phagosomes by counteracting the activity of bacterial ubiquitin E3 ligases.

    1. Computational and Systems Biology
    2. Neuroscience

    Bayesian inference for biophysical neuron models enables stimulus optimization for retinal neuroprosthetics

    Jonathan Oesterle, Christian Behrens ... Philipp Berens
    Research Article Updated

    While multicompartment models have long been used to study the biophysics of neurons, it is still challenging to infer the parameters of such models from data including uncertainty estimates. Here, we performed Bayesian inference for the parameters of detailed neuron models of a photoreceptor and an OFF- and an ON-cone bipolar cell from the mouse retina based on two-photon imaging data. We obtained multivariate posterior distributions specifying plausible parameter ranges consistent with the data and allowing to identify parameters poorly constrained by the data. To demonstrate the potential of such mechanistic data-driven neuron models, we created a simulation environment for external electrical stimulation of the retina and optimized stimulus waveforms to target OFF- and ON-cone bipolar cells, a current major problem of retinal neuroprosthetics.