Inferring causal molecular networks

Empirical assessment through a community-based effort

HPN-DREAM Consortium

Research output: Contribution to journalArticle

Abstract

It remains unclear whether causal, rather than merely correlational, relationships in molecular networks can be inferred in complex biological settings. Here we describe the HPN-DREAM network inference challenge, which focused on learning causal influences in signaling networks. We used phosphoprotein data from cancer cell lines as well as in silico data from a nonlinear dynamical model. using the phosphoprotein data, we scored more than 2,000 networks submitted by challenge participants. The networks spanned 32 biological contexts and were scored in terms of causal validity with respect to unseen interventional data. A number of approaches were effective, and incorporating known biology was generally advantageous. Additional sub-challenges considered time-course prediction and visualization. our results suggest that learning causal relationships may be feasible in complex settings such as disease states. Furthermore, our scoring approach provides a practical way to empirically assess inferred molecular networks in a causal sense.

Original languageEnglish (US)
Pages (from-to)310-322
Number of pages13
JournalNature Methods
Volume13
Issue number4
DOIs
StatePublished - Mar 30 2016

Fingerprint

Phosphoproteins
Learning
Nonlinear Dynamics
Computer Simulation
Visualization
Cells
Cell Line
Neoplasms

ASJC Scopus subject areas

  • Biotechnology
  • Molecular Biology
  • Biochemistry
  • Cell Biology

Cite this

Inferring causal molecular networks : Empirical assessment through a community-based effort. / HPN-DREAM Consortium.

In: Nature Methods, Vol. 13, No. 4, 30.03.2016, p. 310-322.

Research output: Contribution to journalArticle

@article{e7085487fd294d65a9baedb1c9daf243,
title = "Inferring causal molecular networks: Empirical assessment through a community-based effort",
abstract = "It remains unclear whether causal, rather than merely correlational, relationships in molecular networks can be inferred in complex biological settings. Here we describe the HPN-DREAM network inference challenge, which focused on learning causal influences in signaling networks. We used phosphoprotein data from cancer cell lines as well as in silico data from a nonlinear dynamical model. using the phosphoprotein data, we scored more than 2,000 networks submitted by challenge participants. The networks spanned 32 biological contexts and were scored in terms of causal validity with respect to unseen interventional data. A number of approaches were effective, and incorporating known biology was generally advantageous. Additional sub-challenges considered time-course prediction and visualization. our results suggest that learning causal relationships may be feasible in complex settings such as disease states. Furthermore, our scoring approach provides a practical way to empirically assess inferred molecular networks in a causal sense.",
author = "{HPN-DREAM Consortium} and Hill, {Steven M.} and Heiser, {Laura M.} and Thomas Cokelaer and Michael Linger and Nesser, {Nicole K.} and Carlin, {Daniel E.} and Yang Zhang and Artem Sokolov and Paull, {Evan O.} and Wong, {Chris K.} and Kiley Graim and Adrian Bivol and Haizhou Wang and Fan Zhu and Bahman Afsari and Danilova, {Ludmila V.} and Favorov, {Alexander V.} and Lee, {Wai Shing} and Dane Taylor and Hu, {Chenyue W.} and Long, {Byron L.} and Noren, {David P.} and Bisberg, {Alexander J.} and Mills, {Gordon B.} and Gray, {Joe W.} and Michael Kellen and Thea Norman and Stephen Friend and Qutub, {Amina A.} and Fertig, {Elana J.} and Yuanfang Guan and Mingzhou Song and Stuart, {Joshua M.} and Spellman, {Paul T.} and Heinz Koeppl and Gustavo Stolovitzky and Julio Saez-Rodriguez and Sach Mukherjee and Bahman Afsari and Rami Al-Ouran and Bernat Anton and Tomasz Arodz and Sichani, {Omid Askari} and Neda Bagheri and Noah Berlow and Bisberg, {Alexander J.} and Richard Bonneau and Anwesha Bohler and Jaume Bonet and Richard Bonneau",
year = "2016",
month = "3",
day = "30",
doi = "10.1038/nmeth.3773",
language = "English (US)",
volume = "13",
pages = "310--322",
journal = "Nature Methods",
issn = "1548-7091",
publisher = "Public Library of Science",
number = "4",

}

TY - JOUR

T1 - Inferring causal molecular networks

T2 - Empirical assessment through a community-based effort

AU - HPN-DREAM Consortium

AU - Hill, Steven M.

AU - Heiser, Laura M.

AU - Cokelaer, Thomas

AU - Linger, Michael

AU - Nesser, Nicole K.

AU - Carlin, Daniel E.

AU - Zhang, Yang

AU - Sokolov, Artem

AU - Paull, Evan O.

AU - Wong, Chris K.

AU - Graim, Kiley

AU - Bivol, Adrian

AU - Wang, Haizhou

AU - Zhu, Fan

AU - Afsari, Bahman

AU - Danilova, Ludmila V.

AU - Favorov, Alexander V.

AU - Lee, Wai Shing

AU - Taylor, Dane

AU - Hu, Chenyue W.

AU - Long, Byron L.

AU - Noren, David P.

AU - Bisberg, Alexander J.

AU - Mills, Gordon B.

AU - Gray, Joe W.

AU - Kellen, Michael

AU - Norman, Thea

AU - Friend, Stephen

AU - Qutub, Amina A.

AU - Fertig, Elana J.

AU - Guan, Yuanfang

AU - Song, Mingzhou

AU - Stuart, Joshua M.

AU - Spellman, Paul T.

AU - Koeppl, Heinz

AU - Stolovitzky, Gustavo

AU - Saez-Rodriguez, Julio

AU - Mukherjee, Sach

AU - Afsari, Bahman

AU - Al-Ouran, Rami

AU - Anton, Bernat

AU - Arodz, Tomasz

AU - Sichani, Omid Askari

AU - Bagheri, Neda

AU - Berlow, Noah

AU - Bisberg, Alexander J.

AU - Bonneau, Richard

AU - Bohler, Anwesha

AU - Bonet, Jaume

AU - Bonneau, Richard

PY - 2016/3/30

Y1 - 2016/3/30

N2 - It remains unclear whether causal, rather than merely correlational, relationships in molecular networks can be inferred in complex biological settings. Here we describe the HPN-DREAM network inference challenge, which focused on learning causal influences in signaling networks. We used phosphoprotein data from cancer cell lines as well as in silico data from a nonlinear dynamical model. using the phosphoprotein data, we scored more than 2,000 networks submitted by challenge participants. The networks spanned 32 biological contexts and were scored in terms of causal validity with respect to unseen interventional data. A number of approaches were effective, and incorporating known biology was generally advantageous. Additional sub-challenges considered time-course prediction and visualization. our results suggest that learning causal relationships may be feasible in complex settings such as disease states. Furthermore, our scoring approach provides a practical way to empirically assess inferred molecular networks in a causal sense.

AB - It remains unclear whether causal, rather than merely correlational, relationships in molecular networks can be inferred in complex biological settings. Here we describe the HPN-DREAM network inference challenge, which focused on learning causal influences in signaling networks. We used phosphoprotein data from cancer cell lines as well as in silico data from a nonlinear dynamical model. using the phosphoprotein data, we scored more than 2,000 networks submitted by challenge participants. The networks spanned 32 biological contexts and were scored in terms of causal validity with respect to unseen interventional data. A number of approaches were effective, and incorporating known biology was generally advantageous. Additional sub-challenges considered time-course prediction and visualization. our results suggest that learning causal relationships may be feasible in complex settings such as disease states. Furthermore, our scoring approach provides a practical way to empirically assess inferred molecular networks in a causal sense.

UR - http://www.scopus.com/inward/record.url?scp=84978621488&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84978621488&partnerID=8YFLogxK

U2 - 10.1038/nmeth.3773

DO - 10.1038/nmeth.3773

M3 - Article

VL - 13

SP - 310

EP - 322

JO - Nature Methods

JF - Nature Methods

SN - 1548-7091

IS - 4

ER -