Learning causal networks with latent variables from multivariate information in genomic data

Verny, Louis; Sella, Nadir; Affeldt, Séverine; Singh, Param Priya; Isambert, Hervé

doi:10.1371/journal.pcbi.1005662

Cited by 35 publications

(56 citation statements)

References 68 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To distinguish between direct and indirect interactions, several network construction tools use a probabilistic graphical model Kurtz et al (2015); Yang et al (2017), e.g. SPIEC- EASI Kurtz et al (2019, miic Verny et al (2017), or FlashWeave Tackmann et al (2019. FlashWeave can also integrate metadata to remove indirect associations driven by environmental factors.…”

Section: Removing Indirect Dependencies Including Environmental Effectsmentioning

confidence: 99%

Disentangling environmental effects in microbial association networks

Deutschmann

Lima‐Mendez

Krabberød

et al. 2020

Preprint

View full text Add to dashboard Cite

Background: Ecolocial interctions among microorganisms are fundamental for ecosystem function, yet they are mostly unknown or poorly understood. High-throughput-omics can indicate microbial interactions by associations across time and space, which can be represented as association networks. Links in these networks could result from either ecological interactions between microorganisms, or from environmental selection, where the association is environmentally-driven. Therefore, before downstream analysis and interpretation, we need to distinguish the nature of the association, particularly if it is due to environmental selection or not.Results: We present EnDED (Environmentally-Driven Edge Detection), an implementation of four approaches as well as their combination to predict which links between microorganisms in an association network are environmentally-driven. The four approaches are Sign Pattern, Overlap, Interaction Information, and Data Processing Inequality. We tested EnDED on networks from simulated data of 50 microorganisms. The networks contained on average 50 nodes and 1,087 edges, of which 60 were true interactions but 1,026 false associations (i.e. environmentally-driven or due to chance). Applying each method individually, we detected a moderate to high number of environmentally-driven edges—87% Sign Pattern and Overlap, 67% Interaction Information, and 44% Data Processing Inequality. Combining these methods in an intersection approach resulted in retaining more interactions, both true and false (32% of environmentally-driven associations). The addition of noise to the simulated datasets did not alter qualitatively these results. After validation with the simulated datasets, we applied EnDED on a marine microbial network inferred from 10 years of monthly observations of microbial-plankton abundance. The intersection combination predicted that 14.2% of the associations were environmentally-driven, while individual methods predicted 31.4% (Data Processing Inequality), 38.3% (Interaction Information), and up to 83.4% (Sign Pattern as well as Overlap).Conclusions: To reach accurate hypotheses about ecological interactions, it is important to determine, quantify, and remove environmentally-driven associations in marine microbial association networks. For that, EnDED offers up to four individual methods as well as their combination. However, especially for the intersection combination, we suggest to use EnDED with other strategies to reduce the number of false associations and consequently the number of potential interaction hypotheses.

show abstract

Section: Removing Indirect Dependencies Including Environmental Effectsmentioning

confidence: 99%

Disentangling environmental effects in microbial association networks

Deutschmann

Lima‐Mendez

Krabberød

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…As the Set 3 genes specified the perivascular population corresponding to the major cell cluster A we took advantage of the large number of Leptin receptor-positive (LEPR) single cells to refine the V gene network. To identify direct paths between genes including causal relationships and inferring latent common regulators of expressed genes we applied the multivariate information-based inductive causation algorithm (miic) (Sella et al, 2018;Verny et al, 2017), submitting to it the matrix consisting of 1,712 observations and 109 variables (the Set 3 genes). As expected only a fraction of nodes (52) defining direct high-confidence paths were retained ( Figure 4C).…”

Section: Single-cell-level Analysis Identifies High-confidence Pathsmentioning

confidence: 99%

“…We applied a discriminant analysis pipeline to identify the genes predominantly expressed by the different cell sets and analyzed the different cell populations to unravel the gene relatedness (Horvath, 2011;Langfelder and Horvath, 2008). Moreover, we took advantage of the numerous observations provided by recently published single-cell transcriptomes (Tikhonova et al, 2019) to disclose in the perivascular network the direct gene-to-gene interactions and reveal oriented links based on the signature of causality (Sella et al, 2018;Verny et al, 2017). Finally, as proof of concept, we studied the protein expression and activity of the top most discriminant and connected gene, the Wnt signaling pathway facilitator R-spondin 2 (Rspo2), and validated that it was expressed by a small set of perivascular cells, and acted in concert with the Stem Cell Factor (SCF) (also known as Kit Ligand) to amplify ex vivo hematopoietic precursors.…”

Section: Introductionmentioning

confidence: 99%

Inferring Gene Networks in Bone Marrow Hematopoietic Stem Cell-Supporting Stromal Niche Populations

Desterke

Petit²,

Sella³

et al. 2020

iScience

Self Cite

View full text Add to dashboard Cite

The cardinal property of bone marrow (BM) stromal cells is their capacity to contribute to hematopoietic stem cell (HSC) niches by providing mediators assisting HSC functions. In this study we first contrasted transcriptomes of stromal cells at different developmental stages and then included large number of HSC-supportive and non-supportive samples. Application of a combination of algorithms, comprising one identifying reliable paths and potential causative relationships in complex systems, revealed gene networks characteristic of the BM stromal HSCsupportive capacity and of defined niche populations of perivascular cells, osteoblasts, and mesenchymal stromal cells. Inclusion of single-cell transcriptomes enabled establishing for the perivascular cell subset a partially oriented graph of direct gene-to-gene interactions. As proof of concept we showed that R-spondin-2, expressed by the perivascular subset, synergized with Kit ligand to amplify ex vivo hematopoietic precursors. This study by identifying classifiers and hubs constitutes a resource to unravel candidate BM stromal mediators. Recent transcriptomic and proteomic analyses at the single-cell level have helped define the gene sets characterizing the BM stromal populations under steady state, after stress, or in diseased conditions (Bal

show abstract

“…Gene ontology analysis was done using the human genome as background (Table S1). Functional gene network reconstruction was achieved using the information-theoretic method, MIIC (multivariate information-based inductive causation, Supplementary Information 1 [58,65]). Four independent datasets comprising 153 to 485 primary GBM transcriptomes were used for patient survival analysis (Table S1).…”

Section: Computational Analysesmentioning

confidence: 99%

Capture at the single cell level of metabolic modules distinguishing aggressive and indolent glioblastoma cells

Saurty

Bellenger

El-Habr

et al. 2019

Preprint

View full text Add to dashboard Cite

Glioblastoma cell ability to adapt their functioning to microenvironment changes is a source of the extensive intra-tumor heterogeneity characteristic of this devastating malignant brain tumor. A systemic view of the metabolic pathways underlying glioblastoma cell functioning states is lacking.We analyzed public single cell RNA-sequencing data from glioblastoma surgical resections, which offer the closest available view of tumor cell heterogeneity as encountered at the time of patients' diagnosis. Unsupervised analyses revealed that information dispersed throughout the cell transcript repertoires encoded the identity of each tumor and masked information related to cell functioning states. Data reduction based on an experimentally-defined signature of transcription factors overcame this hurdle. It allowed cell grouping according to their tumorigenic potential, regardless of their tumor of origin. The approach relevance was validated using an independent dataset of glioblastoma tissue transcriptomes, patient-derived cell lines and orthotopic xenografts.Overexpression of genes coding for amino acid and lipid metabolism enzymes involved in antioxidative, energetic and cell membrane processes characterized cells with high tumorigenic potential. Modeling of their expression network highlighted the very long chain polyunsaturated fatty acid synthesis pathway at the core of the network. Expression of its most downstream enzymatic component, ELOVL2, was associated with worsened patient survival, and required for cell tumorigenic properties in vivo. Our results demonstrate the power of signature-driven analyses of single cell transcriptomes to obtain an integrated view of metabolic pathways at play within the heterogeneous cell landscape of patient tumors.

show abstract

Learning causal networks with latent variables from multivariate information in genomic data

Cited by 35 publications

References 68 publications

Disentangling environmental effects in microbial association networks

Disentangling environmental effects in microbial association networks

Inferring Gene Networks in Bone Marrow Hematopoietic Stem Cell-Supporting Stromal Niche Populations

Capture at the single cell level of metabolic modules distinguishing aggressive and indolent glioblastoma cells

Contact Info

Product

Resources

About