Biologie computationnelle

Mitochondrial computational biology

Unravelling mitochondrial adaptability with the mitoXplorer platform

Mitochondrial structure and content differs from one cell type to the other. Mitochondrial variability is also at the heart of many disease conditions, often related to the nervous system or the muscle.

We are interested in understanding the dynamics of mitochondria in different species, tissues, disease states and throughout ageing. We look at the expression dynamics – reflected in systematic transcriptomic and proteomic data – from a variety of different tissues, disease states, ages and in 4 different species to get an idea of what portion of mito-genes are dynamically expressed.

We have developed the mitoXplorer platform, an interactive, visual data mining and integration platform to explore large-scale expression and mutation data from a mitochondrial perspective.

A manually curated, mitochondrial interactome is at the heart of mitoXplorer, which contains all proteins that have mitochondrial function, irrespective of their genomic localization (mitochondrial or nuclear gnome). We have annotated the mito-interactome with 38 mitochondrial processes. We provide currently the most up-to-date and complete, high-quality manually curated and annotated interactomes for 4 species: human, mouse, Drosophila melanogaster and Saccharomyces cerevisiae.

We integrate these manually curated and annotated mito-interactomes with -omics data from large public repositories, such as the Cancer Genome Atlas (TCGA) or with data from Gene Expression Omnibus (GEO). Using specialised pipelines for NGS-data analysis, we extract mutation and expression data for all mito-proteins.

mitoXplorer offers 7 highly interactive data analysis and visualisation tools to explore the dynamics of mito-associated expression changes and mutations in a variety of data sets from different experimental or disease conditions.

Moreover, 2 integrative interfaces allow us to search for transcriptional, as well as signalling regulators of mitochondrial dynamics. This enables us to analyse, visualise and integrate a large variety of -omics data-sets from a mitochondrial perspective and learn how mitochondria adapt to cellular changes.

Check out mitoXplorer here: http://mitoxplorer2.ibdm.univ-mrs.fr

Download your own version of mitoXplorer here: https://gitlab.com/habermann_lab/MitoX2

The update of mitoXplorer is now published in NAR: https://doi.org/10.1093/nar/gkac306

This project was supported by DFG grant ‘CancerSysDB’ and is currently supported by ANR grant ‘MITO-DYNAMICS’, the FRM grant ‘ataxiaXplorer’, and the CNRS.

Mitochondria in neurodegenerative diseases: the ataxiaXplorer platform

In an FRM-funded project we hold together with the team of Helene Puccio (INMG Lyon), we are creating the first sister platform of mitoxplorer, ataxiaXplorer. AtaxiaXplorer will be based on the mitoXplorer 2.0 web-server, with an ataxia-specific interactome and enhanced functionalities to mine single-cell RNA-seq data.

Annotating and integrating epigenetic and transcription factor data using AnnoMiner

The transcriptional regulation of genes is a highly complex process, requiring the precisely timed activation of transcription factors (TFs), enhancers or repressors. We still don’t completely understand this process, though recent technological advances provide more and deeper insight into the mechanisms involved in transcriptional regulation. These include techniques to observe TF/enhancer binding to DNA, modification of histones and DNA, opening or closing of the DNA-structure, as well as larger-scale structural re-organization of the DNA.

One key step for computational biologists is to analyse and interpret these epigenetic and TF occupancy data. This includes the annotation of these epigenetic and occupancy data with genomic features, with each other, as well as with available data on gene expression dynamics.

With AnnoMiner, we provide a web-platform that allows us to annotate and integrate epigenetic and TF occupancy data with genomic features and with each other. Optionally, the user can also integrate gene expression data, enabling the identification of potential direct targets of on, or of a combination of transcriptional regulators. Gene features are divided into upstream/downstream regions, transcription start site (TSS) untranslated regions, as well as the coding region; from a graphical ‘coverage plot’ that shows the overlap of a transcriptional regulator under study with these regions, the user can choose which region is most interesting to him for subsequent annotation.

AnnoMiner also allows to mine 10 nearby genes to a TF peak of genomic variant, with the nearby gene function. For this function, a single peak or region can be submitted, in combination with a list of differentially expressed genes. The 5 upstream and downstream genes to the peak will be visualised, together with information on their deregulation.

Finally, AnnoMiner can be used to annotate a list of differentially expressed genes to look for transcription factors whose peaks are enriched in the submitted list.

Check out AnnoMiner here: http://chimborazo.ibdm.univ-mrs.fr/AnnoMiner
AnnoMiner has been published here: https://doi.org/10.1038/s41598-021-94805-1

Network Biology

Biological networks such as protein-protein interaction networks or gene regulatory networks are an integral part to understanding biological systems. Yet, our current representations of biological networks are mostly static: the factor of time (for instance coming from time-series data) or event-driven interaction (as is found in cell cycle regulation) are not accounted for.

In collaboration with groups from the CPT (Alain Barrat) and I2M (Laurent Tichit), we are integrating time as a factor in biological networks.

Using higher-order networks to model temporal biological systems

Many biological systems go through various phases, or states, over different time and space scales. These phases, and their order in time, are often crucial to the functioning of these systems and even determine their evolution/fate. To understand or even predict these phases would yield crucial insight into the temporal organisation of such systems, and ultimately, into their evolution.

The cell cycle, at the heart of all biological development, illustrates this well. Indeed, in order to eventually divide into two cells, the cell typically progresses through 4 macroscopic phases. At the microscopic level, these phases are driven by protein-protein interactions (PPIs).

To investigate the temporal organisation of the cell cycle, however, the typical static PPI network is not enough. That is why we build a temporal PPI network, by integrating time series of protein concentrations to the static one. By doing so, we open the whole toolbox of temporal network and higher-order networks to study our biological system.

We infer cell cycle phases and subphases, revealing the temporal organisation of the cycle over multiple timescales. We do this by clustering snapshots of the temporal network. This project is funded by CENTURI and is done in collaboration with Alain Barrat (CPT) and Laurent Tichit (I2M)

TimeNexus: a cytoscape app to analyse time-series expression data with temporal multilayer networks

Biological networks are not static, but dynamic. With the availability of time-series datasets, we can follow the dynamics of expression changes – on protein or RNA level – experimentally. To integrate these types of data with biological networks is currently not so easy, as few ready-to-use tools exist. In order to address this gap, we have developed the Cytoscape app TimeNexus. In TimeNexus, we project temporal data on a multilayer network: each layer represents one time-point. We refer to these networks as temporal multilayer networks (tMLNs).

With TimeNexus, a user can easily create such a tMLN simply by providing tabular information on the network itself, as well as the temporal layers in form of time-series data. TimeNexus furthermore provides a framework for visualising a tMLN, as well as for extracting active subnetworks using the Cytoscape apps PathLinker and the ANAT server.

Take a look at the TimeNexus paper: https://doi.org/10.1038/s41598-021-93128-5
You can also download the TimeNexus app directly from the Cytoscape App store: https://apps.cytoscape.org/apps/timenexus

This project is done in close collaboration with Laurent Tichit (I2M). It was funded by a shared ANR grant our team has with Aziz Moqrich (IBDM), ‘MYOCHRONIC’.

Evolutionary Computational Biology

The process of evolution is universal and never-ending. It allows organisms to adapt to their environment, can create novel function, but can also lead to deleterious loss. Computationally, as well as experimentally, evolution is studied in many ways and from many different angles.

Evolution of protein short linear motifs

Short linear sequence motifs in proteins are essential for proteins to interact with each other and other (macro-)molecules. While we know millions of proteins, our knowledge of protein motifs is rather limited: the Eukaryotic Linear Motif (ELM) database lists to date only ~5000 known and experimentally verified protein motifs. Experimentally, it is labour- and time-intensive to identify novel protein motifs. Computationally, the task is equally difficult, due to the shortness and generally low conservation of motifs. Novel experimental techniques (mostly based on Mass-Spec) have uncovered a broader range of functional motifs that we can use for studying them.

Protein motifs are also interesting in the light of evolution: by motif gain or loss, protein function can be easily modified: as protein motifs are very short – and are thought to reside primarily in disordered regions of proteins, ex nihilo motif evolution can lead to novel protein functions.
We are interested in understanding motif evolution. Depending on their localization in the protein, they might experience different evolutionary pressures: a protein motif in a disordered region might evolve more easily than one in a fully structured region.

We look at motif evolution in a structural context and do so in a comprehensive manner, considering all so-far known protein motifs that are experimentally verified. We want to understand how motifs evolve in different structural contexts and identify novel motifs that lead to phenotypic changes of their ‘host’ protein. To achieve this goal, we are developing computational pipelines to put protein motifs in their structural, as well as evolutionary context.

Mechano-sensing motifs in proteins

In the protein motif world, we are also interested in finding motifs that render proteins mechano-sensitive. Mechano-sensitivity is an important trait required to react to mechanical signals – or stress – in a cell. We need mechano-sensitivity for instance to contract our muscle cells. Many examples of mechanical signals or stress exist, which changes the behaviour of our cells.
In a project which is carried out in very close collaboration with the team of Felix Rico (LAI), we look at the nature of protein motifs that react to mechanical stress. We do so by combining structural techniques, machine-learning as well as computational techniques.

Evolution of epithelia

Sponges (Porifera) are at the base of animal evolution; they are some of the first, metazoan (multicellular) animals. Sponges are also popular model organisms to study early traits of multicellular animals. One such trait is the presence of an epithelium, which helps define an ‘inner’ and an ‘outer’ side of a body or an organ.

Using the homoscleromorph sponge Oscarella lobularis and in close collaboration with the labs of Andre le Bivic (IBDM) and Carole Borchiellini (IMBE), we are looking at the evolution of epithelia in these animals.
In homoscleromorph sponges, we can find all features of a true epithelial structure, including adherens junctions, a basement membrane and cellular polarity. Using bulk RNA-sequencing of disaggregation / reaggregation experiments in O. lobularis, as well as single-cell RNA sequencing of O. lobularis adult animals, as well as (asexually produced) buds, we aim to unravel the molecular structure of the sponge epithelial toolkit and thus, to take a closer look at the evolution of epithelia themselves.

We also currently work on an improved transcriptome of O. lobularis, as well as its complete genome sequence – assembled at contig level and sequenced with long-read sequencing – hopefully coming soon.

Evolution of predatory traits

In evolution, the Red Queen Hypothesis states that species must constantly adapt, evolve and proliferate to survive in the constant battle against co-evolving species: the prey needs to evolve physical traits (features) that allow it to escape the predator; this in turn will induce the predator to evolve better skills to catch the prey. Many examples exist in the animal kingdom that demonstrate this co-evolutionary process. In animals this can mean prey becomes faster or camouflages better or evolves other defensive strategies to avoid being killed and eaten by a predator.In the bacterial world, there are also predators that kill and live from other bacteria, archaea or even fungi. Myxococcus xanthus is a famous example, which kills and consumes other such species. M. xanthus is a social bacterium. It hunts in the group (wolf-pack hunting) and has a complex life cycle, depending on the availability of prey. However, M. xanthus will not be similarly efficient with all types of prey, as prey-bacteria will also have different defence mechanisms at their hand. While it can easily eat E. coli, it might not be so successful with other types of bacteria.

We look at predator-prey evolution in the simpler bacterial world. In a very close collaboration with the team of Tam Mignot (from LCB, Marseille), we investigate how the predatory bacteria Myxococcus xanthus and its prey, E. coli, co-evolve and develop higher predatory or higher defence skills. We combine our experiments with -omics techniques to identify causative and adaptive mutations.

When we subject prey to M. xanthus, what type of defence mechanisms will it evolve? When we subject M. xanthus to resistant prey, will it evolve better hunting abilities? In order to address this, we combine co-evolution experiments of predator-prey for 100s of generations with genome and transcriptome sequencing to identify the molecular players involved in predator and prey efficiency.

Computational tool development for biological data mining and integration

Over the years, we have developed a number of user-friendly tools tailored for wet-lab biologists to analyse, mine, visualise and integrate biological data of different origin. Most of them are still up and running. Check out our collection at this page. We tried to order them by theme.

Omics data integration and visualisation

mitoXplorer

mitoXplorer is a web-tool for the analysis, visualisation and integration of -omics data in a mitochondria-centric way.

Web-link: http://mitoxplorer2.ibdm.univ-mrs.fr

Tutorial: http://mitoxplorer2.ibdm.univ-mrs.fr/tutorials.html

Source-code: https://gitlab.com/habermann_lab/MitoX2

Publication: Marchiano, F, Haering M and Habermann BH. The mitoXplorer 2.0 update: integrating and interpreting mitochondrial expression dynamics within a cellular context. NAR web server issue 2022 (https://doi.org/10.1093/nar/gkac306).

AnnoMiner

AnnoMiner allows annotation of ChIP-seq (epigenetic, TF-binding) data and to perform enrichment analysis using the ENCODE/modERN resources.

Web-link: http://chimborazo.ibdm.univ-mrs.fr/AnnoMiner/

Tutorial: http://chimborazo.ibdm.univ-mrs.fr/AnnoMiner/tutorial.html

Source-code: https://gitlab.com/habermann_lab/AnnoMiner

Publication: Meiler A, Marchiano F, Haering M, Weitkunat M, Schnorrer F, Habermann BH. AnnoMiner is a new web-tool to integrate epigenetics, transcription factor occupancy and transcriptomics data to predict transcriptional regulators. Sci Rep. 2021 Jul 29;11(1):15463. doi: 10.1038/s41598-021-94805-1.

RNfuzzyApp

RNfuzzyApp is a R-shiny app for the differential expression and time-course analysis of RNA-seq data.

Tutorial: https://gitlab.com/habermann_lab/rna-seq-analysis-app/-/blob/master/RNfuzzyApp_UserManual.pdf

Source-code: https://gitlab.com/habermann_lab/rna-seq-analysis-app

Publication: Haering M and Habermann BH. RNfuzzyApp: an R shiny RNA-seq data analysis app for visualisation, differential expression analysis, time-series clustering and enrichment analysis [version 2; peer review: 3 approved]. F1000Research 2021, 10:654, PMID: 35186266, https://doi.org/10.12688/f1000research.54533.2.

SLALOM

SLALOM is a suite of python programs for statistical analysis based on positional data in proteins and genomes.

Source-code: https://gitlab.com/habermannlab/slalom

Publication: Prytuliak R, Pfeiffer F, Habermann BH. SLALOM, a flexible method for the identification and statistical analysis of overlapping continuous sequence elements in sequence- and time-series data. BMC Bioinformatics. 2018 Jan 26;19(1):24. doi: 10.1186/s12859-018-2020-x.

Tools for network analysis in biology

TimeNexus

TimeNexus is a Cytoscape app that allows users to extract ‘active’ pathways across a temporal expression data set by using temporal multilayer networks.

Tutorial: https://gitlab.com/habermann_lab/timenexus-cytoscape-app

Source-code: https://apps.cytoscape.org/apps/timenexus

Publication: Pierrelée, M., Reynders, A., Lopez, F., Moqrich, A., Tichit, L., Habermann, B.H. Introducing the novel Cytoscape app TimeNexus to analyse time-series data using temporal MultiLayer Networks (tMLNs). Sci Rep 11, 13691 (2021). https://doi.org/10.1038/s41598-021-93128-5.

Phasik

Phasik is a novel method based on temporal network clustering to extract phases from temporal, biological data (such as time-course expression data).

Tutorial: https://phasik.readthedocs.io/en/latest/tutorial/index.html

Notebooks: https://gitlab.com/habermann_lab/phasik/-/tree/master/notebooks

Source-code: https://gitlab.com/habermann_lab/phasik

Publication: Maxime Lucas, Arthur Morris, Alex Townsend-Teague, Laurent Tichit, Bianca H. Habermann, Alain Barrat, doi: https://doi.org/10.1101/2021.03.26.437187

viPEr & PEANuT

viPEr and PEANuT are two Cytoscape 3 apps for focus network analysis.

Source-code: https://gitlab.com/habermann_lab/viper-virtual-pathway-explorer

Publication: Garmhausen M, Hofmann F, Senderov V, Thomas M, Kandel BA, Habermann BH. Virtual pathway explorer (viPEr) and pathway enrichment analysis tool (PEANuT): creating and analyzing focus networks to identify cross-talk between molecules and pathways. BMC Genomics. 2015 Oct 14;16:790. https://doi.org/10.1186/s12864-015-2017-z.

Finding and analysing short linear motifs in proteins

HH-MOTiF

HH-MOTiF finds short linear motifs (SLiMs) de novo in proteins using Hidden Markov Model (HMM) comparisons.

Web-link: http://hh-motif.ibdm.univ-mrs.fr/search/

Source-code: https://framagit.org/BCF/hhmotif

Publication: Prytuliak R, Volkmer M, Meier M, Habermann BH. HH-MOTiF: de novo detection of short linear motifs in proteins by Hidden Markov Model comparisons. Nucleic Acids Res. 2017 Jul 3;45(W1):W470-W477. doi: 10.1093/nar/gkx341.

Evo-motif

evo-MOTIF places protein short linear motifs in their evolutionary and structural context.

Source-code: https://gitlab.com/habermann_lab/slims

Finding orthologs or homologs

MorFeus

morFeus finds remotely conserved orthologs based on iterative, relaxed, reciprocal BLAST searches and network scoring

Web-link: http://morfeus.ibdm.univ-mrs.fr/

Source-code: https://framagit.org/BCF/morfeus

Publication: Wagner I, Volkmer M, Sharan M, Villaveces JM, Oswald F, Surendranath V, Habermann BH. morFeus: a web-based program to detect remotely conserved orthologs using symmetrical best hits and orthology network scoring. BMC Bioinformatics. 2014 Aug 6;15:263. doi: 10.1186/1471-2105-15-263.

HMMerThread

HMMerThread is a data-resource of remotely conserved domains in protein sequences.

Web-link: https://hmmerthread.gurdon.cam.ac.uk/cgi-bin/HMMerSearch.py

Publication: Bradshaw CR, Surendranath V, Henschel R, Mueller MS, Habermann BH. HMMerThread: detecting remote, functional conserved domains in entire genomes by combining relaxed sequence-database searches with fold recognition. PLoS One. 2011 Mar 10;6(3):e17568. https://doi.org/10.1371/journal.pone.0017568

Membres de l'équipe

Le groupe de biologie computationnelle aborde les problèmes biologiques à l'aide de méthodes computationnelles, en se concentrant sur les mitochondries en tant qu'organites métaboliques importants et sur l'évolution des systèmes biologiques.

Publications

Actualités