# Christophe Pouzat

## Table of Contents

CNRS UMR 8145

45, rue des Saints Pères

75270 Paris Cedex 06

France

`christophe.pouzat-at-parisdescartes.fr`

`christophe.pouzat-at-gmail.com`

Cette page en français.

## Background

- I am currently a CNRS researcher of the MAP5 the Applied Maths Laboratory of the Paris-Descartes University.
- I have been for 11 years (Jan 2001 - Dec 2011) member of the Laboratory of Brain Physiology of the same University.
- I did a two years post-doc (Jan 1999 - Dec 2000) in the laboratory of Gilles Laurent at the California Institute of Technology.
- I did my PhD at the Max Planck Institute for Biophysical Chemistry under the supervision of Alain Marty (Feb 1995 - Dec 1998).

## Research

Having a background in *experimental* neurophysiology, I mainly work on the analysis of data produced in this field. This work can be split into three broad categories – with associated keywords linking the task to specific statistical methodologies :

- Spike sorting – dimension-reduction, clustering, classification, Gaussian mixture models, EM algorithm, MCMC.
**All that is now explained in a short film on YouTube**. - Spike train analysis – point process / counting process, conditional intensity estimation, nonparametric estimation, smoothing spline, goodness of fit tests, Donsker theorem.
- Calcium imaging – Poisson regression, variance stabilization, parametric / nonparametric models, image segmentation.

In every project I try to implement what is now commonly called the reproducible research paradigm – that I prefer to call the *reproducible data analysis* paradigm. I am not a developer of the tools making the implementation of the paradigm possible but a daily user of them.

## Collaborations

**Within Paris-Descartes University**I'm collaborating with:- Yves Rozenholc on the development of nonparametric methods for the analysis of fluorescence calcium measurements.
- Avner Bar-Hen on the development of goodness of fit tests for counting process models.

**In the rest of the World**I'm collaborating with:- Peter Kloppenburg from Cologne University (Germany), my main "data provider".
- Chong Gu from Purdue University (USA) with whom I'm developing smoothing spline based approaches to estimate the intensity of the counting processes used to model spike trains.
- Clément Léna of the Cerebellum Group (Biology Department, École Normale Supérieure, Paris) for the analysis of MEA recording performed
*in vivo*in the rodent cerebellum. - Antoine Chaillet and Georgios Detorakis (Supélec) and Yann-Suhan Senova (Équipe d'accueil 4391 - Excitabilité nerveuse et thérapeutique, Henri Mondor Hospital, Créteil, France) on spike sorting and the analysis of spike trains of data recorded
*in vivo*from primates : The SynchNeuro ANR project ; - Patricia Reynaud-Bouret, Christine Tuleau and Frank Grammont (J. A. Dieudonné Laboratory, University of Nice Sophia-Antipolis) on the analysis of spike trains ;
- Antonio Galves (NeuroMat at the University of São Paulo) on spike sorting and spike train analysis.

## Research animation

- I am a member of the "guiding committee" of the Multi-electrodes Research Group.
- I am a member of the organizing committee of the Semin-R, a Parisian R users' group.

## Software

**Software I use**: I could not work without Arch or Manjaro Linux, emacs + Org + babel, Python, R, maxima, SBCL, conkeror, TeX Live.**Software I develop**:- My current efforts on spike sorting can be found and downloaded from my PouzatDetorakisEuroScipy2014 on GitHub or zenodo (doi: 10.5281/zenodo.15070) and my Neuronal-spike-sorting github as well as my Spike sorting pages.
- STAR (Spike Train Analysis with R) is an R package.
- CalciOMatic an R package for fluorescence calcium measurements analysis, developed jointly with Sébastien Joucla.
- DynClust an R package for non-parametric denoising and clustering of noisy images indexed by time, developed jointly with Tiffany Lieury and Yves Rozenholc.

## Publications

My five latest publications:

- Christophe Pouzat and Patricia Reynaud-Bouret (2016) Editorial for the special issue on Statistics and Neurosciences
*Journal de la Société Française de Statistique***157(1)**: 1-2. The pdf is available. - Christophe Pouzat, Andrew Davison and Konrad Hinsen (2015) La recherche reproductible : une communication scientifique explicite
*Statistique et Société*(in French)**3(1)**: 35-38. The pdf and some additional marterial are available. - Christophe Pouzat and Georgios Is Detorakis (2014) SPySort: Neuronal Spike Sorting with Python
*Proceedings of the 7th European Conference on Python in Science (EuroSciPy 2014)*(web site), edited by Pierre de Buyl and Nelle Varoquaux, pp. 27–34 (arXiv:1412.6383). A "detailed" version hal-01111654v1 with all the codes is available, as well as its`org-mode`

source file both on GitHub and on zenodo (doi: 10.5281/zenodo.15070). - Sébastien Joucla, Romain Franconville, Andreas Pippow, Peter Kloppenburg and Christophe Pouzat (2013) Estimating background-subtracted fluorescence transients in calcium imaging experiments: a quantitative approach
*Cell Calcium***54**(2): 71-85. - Matthieu Delescluse, Romain Franconville, Sébastien Joucla, Tiffany Lieury and Christophe Pouzat (2012) Making neurophysiological data analysis reproducible. Why and how?
*Journal of Physiology (Paris)***106**(3-4): 159-170 (hal, R vignette version of the toy example, odfWeave R vignette version of the toy example, Org/Babel version of the toy example using Maxima and Python, Org/Babel version of the toy example using Common Lisp and gnuplot, BibTeX).

More…

## Pre-prints & reports

- Stephen Eglen, Ben Marwick, Yaroslav Halchenko, Michael Hanke, Shoaib Sufi, Padraig Gleeson, R Angus Silver, Andrew Davison, Linda Lanyon, Mathew Abrams, Thomas Wachtler, David J Willshaw, Christophe Pouzat, Jean-Baptiste Poline (2016) Towards standard practices for sharing computer code and programs in neuroscience (bioRxiv doi: http://dx.doi.org/10.1101/045104).
- Christophe Pouzat, Antoine Chaffiol and Avner Bar-Hen (2015) Homogeneity and identity tests for unidimensional Poisson processes with an application to neurophysiological peri-stimulus time histograms (hal-01113126, version 2) and the companion documents: Supplementary Material for: Homogeneity and identity tests for unidimensional Poisson processes with an application to neurophysiological peri-stimulus time histograms–R version (hal-01224765, version 1) and Supplementary Material for: Homogeneity and identity tests for unidimensional Poisson processes with an application to neurophysiological peri-stimulus time histograms–Python version (hal-01224764, version 1).
- Christophe Pouzat and Georgios Detorakis (2015) A Prelude to SPySort: Neuronal Spike Sorting with Python (hal-01111654, version 1) this is the "supplementary material" of SPySort: Neuronal Spike Sorting with Python (arXiv:1412.6383). The
`org-mode`

source file is available both on GitHub and on zenodo (doi: 10.5281/zenodo.15070)

## Talks

Recent talks:

- Five lectures given at the Latin American School on Computational Neuroscience (LASCON) in January 2016.
- On the relation between neuronal size and extracellular spike amplitude and its consequence on extracellular recordings interpretation. A talk given at the MathStatNeuro workshop, Nice, September 8-10 2015.
- Fluorescence Imaging Analysis: The Case of Calcium Transients (doi: 10.5281/zenodo.18691). A course given at the ENP summer school (June 17 2015). The
`org-mode`

source file—everything is done with Python—is available on GitHub (doi: 10.5281/zenodo.18911). The IPython notebook can be visualized with nbviewer. - Nine Years of Reproducible Research: From R / Sweave to Org given at the INCF Workshop on Encourage Sharing in Cambridge 15-16 December 2014 (org source file).
- Three lectures on spike sorting (also available in video) given at the Research, Innovation and Dissemination Center for Neuromathematics (NeuroMat), Universidade de São Paulo, November 25-27 2014.
- Peri-Stimulus Time Histograms Estimation Through Poisson Regression Without Generalized Linear Models. A talk given a the workshop on mathematical modeling and statistical analysis in neuroscience Copenhagen, July 2-4, 2014 (doi: 10.5281/zenodo.14660). The
`org-mode`

source file is available on GitHub and on zenodo (doi: 10.5281/zenodo.15098). - Nonparametric analysis of simultaneously recorded spike trains considered as a realization of a multivariate point process (doi: 10.5281/zenodo.14713). A talk given at the CANMUM 2014 meeting. The
`org-mode`

source file is available both on GitHub and on zenodo (doi: 10.5281/zenodo.15093). - Des données neurophysiologiques aux modèles statistiques et aux développement de logiciels : retour d'expérience sur l'analyse des séquences de potentiels d'action et l'analyse de données de fluorescence liée au calcium. A talk (in French) given at the deuxième rencontre Math, bio & médecine au sein du PRES Sorbonne Paris Cité, September 19 2013 (org source file).
- From Neurophysiological Data to Statistical Models and Software Development. A talk given at the workshop MATHEMATICS AND NEUROSCIENCE: A DIALOGUE organized by Roberto Fernández at Utrecht University, September 3-5 2013.
- Fluorescence Imaging Analysis: The Case of Calcium Transients (doi: 10.5281/zenodo.14725). A course given at the ENP summer school (June 26 2013). The
`org-mode`

source file is available both on GitHub and on zenodo (doi: 10.5281/zenodo.15097). - Analyse non paramétrique de séquences de potentiels d'action. Construction de modèles et de tests de qualité d'ajustement. (in French) given at the "Premières rencontres R", Bordeaux, July 3rd 2012 (Org source file: this file contains an expended version of the talk; an abstract is available from the conference web site).
- Sept ans de recherche reproductible au quotidien : de R / Sweave à Org (in French) given at the Rencontre de réflexion autour de la recherche reproductible, April 5 2012, Orléans University (Org source file).
- Les séquences de potentiels d’action neuronaux comme observations de processus de comptage : estimation de l’intensité par vraisemblance pénalisée (in French), talk given at the Groupe de travail en Statistiques of the MAP5, June 17 2011 (org source file).
- Spike sorting and sorting quality evaluation through generative model estimation, data simulation, flexible coarse classification and refined template matching. A talk given at the Validation of Automatic Spike-Sorting Methods (Ski, Norway, May 19-20, 2011).
- Making Neurophysiological Data Analysis Reproducible: An Example with the Open Source Software R. Handout of an overview talk given at the Janelia Workshop on Challenges in Extracellular Electrophysiology: Data Extraction (May 16-18, 2010). Spike Sorting: A Reproducible Example using R, a document illustrating the approach discussed in the talk. ReproducibleAnalysis.tar.gz, a bunch of files with everything needed to regenerate the former analysis (using R). In other words, you can check that what I'm describing works as described.
- Spike train measurement and sorting, handout of an overview talk given at Saarbrücken, Universität des Saarlandes, May 7 2010.

More…

## Datasets

The datasets used in my publications are available in HDF5 format from zenodo. You are welcome to do what you want with them–including publishing with them *without having to "invite" me to sign your publication*–just cite your source (the DOI are given bellow):

- Tetrode recording from the antennal lobe of a locust (Schistocerca americana) (doi: 10.5281/zenodo.14607).
- Dataset from "Matthieu Delescluse and Christophe Pouzat (2006) Efficient spike-sorting of multi-state neurons using inter-spike intervals information Journal of Neuroscience Methods 150: 16-29." (doi: 10.5281/zenodo.15228).
- Data set from Pouzat and Chaffiol (2009) Journal of Neuroscience Methods 181:119.(doi: 10.5281/zenodo.14281).
- All my postdoc extracellular recordings from the locust olfactory pathway are now available –a third of the data has been lost due to CD corruption after 14 years!– (doi: 10.5281/zenodo.21589). There are roughly 15 GB of data from 14 experiments.

## Acknowledgments

This web site was edited with emacs + Org. The *css* used comes from the one of Eric Schulte.