1. Fast, Accurate Causal Search Algorithms from the Center for Causal Discovery (CCD)
The CCD Algorithms Group University of Pittsburgh Carnegie Mellon University Pittsburgh Supercomputing Center Yale University
The project is a collaborative effort among investigators at Pitt, UPMC, CMU, and the Pittsburgh Supercomputer Center.
Some of the key personnel on the project have been collaborating on methods for causal modeling and discovery for more than 20 years.
One driving biomedical problem will be the discovery of cell signaling pathways in several cancers, including breast, lung, and colon cancer. Understanding these pathways is central to developing drug therapies that can effectively treat the cancers.
Another driving biomedical problem will be the discovery of the mechanisms of disease in COPD.
Both of these biomedical problems are translational and involve using data that spans from the molecular to the clinical.
BD2K All Hands Meeting
11/29/2016

2. Causal Discovery in Biomedicine
Science is centrally concerned with the discovery of causal relationships in nature
Understanding
Control
Examples:
Determine the genes and cell signaling pathways that cause breast cancer
Discover the clinical effects of a new drug
Uncover the mechanisms of pathogenicity of a recently mutated virus that is spreading rapidly in the population
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

3. Why Establish a Center for Causal Discovery Now?
Algorithmic Advances +
Availability of Big Biomedical Data
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

4. Algorithmic Advances
In the past 25 years, there has been tremendous progress in the development of computational methods for representing and discovering causal networks from a combination of observational data, experimental data, and knowledge.
These methods are generally applicable to biomedical data.
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

5. Availability of Big Biomedical Data
The variety, richness, and quantity of biomedical data have been increasing very rapidly.
The appropriate analysis of these data has great potential to advance biomedical science.
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

6. Primary Goals of the CCD
Goal 1. Develop and implement state-of-the-art methods for discovering causal knowledge from biomedical big data using causal graphical models
Make some of the best existing causal discovery methods available as free, open source software
Develop new methods and make them available
Goal 2. Investigate three biomedical projects (cancer, lung disease, brain functional connectivity) to evaluate methods and drive their further development
Goal 3. Disseminate causal discovery software and knowledge widely to biomedical researchers and data scientists
A group of investigators in the School of Medicine at the University of Pittsburgh will be submitting an application on the topic of modeling and discovery of causal networks from big biomedical data.
The primary aims will be to advance the representation, discovery, and uses of causal network models when applied to very large biomedical datasets.

7. Typical Causal Analysis Workflow
Prior Knowledge
Causal
Analysis
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.
Data
Causal Network

8. Typical Causal Analysis Workflow
Prior Knowledge
Causal
Analysis
Causal Hypothesis Generation by Biomedical Scientists
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.
Data
Causal Network

9. Typical Causal Analysis Workflow
Prior Knowledge
Causal
Analysis
Causal Hypothesis Generation by Biomedical Scientists
Experiments
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.
Data
Causal Network

10. Typical Causal Analysis Workflow
Prior Knowledge
Causal
Analysis
Causal Hypothesis Generation by Biomedical Scientists
Experiments
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.
Data
Causal Network

11. Typical Causal Analysis Workflow
Prior Knowledge
Causal
Analysis
Causal Hypothesis Generation by Biomedical Scientists
Experiments
Data
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.
Causal Network

12. Basic Components Needed to Learn Causal Networks from Data
Model representation
Model evaluation
Model search
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.

13. Model Represenation Causal Bayesian network (CBN)
Directed acyclic graph
Nodes represent variables
Arcs represent causal influence
Specify P(X | parents(X)) for each X
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.
This figure is adapted from: Sachs K, et al. Protein-signaling networks learned from multi-parameter single-cell data of
human T cells Science 308 (2005)

14. Model Representation with CBNs
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.

15. Model Representation Issues
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.

16. Model Evaluation
Constraint based (e.g., tests of conditional independence)
Score based (e.g., Bayesian scores)
Causation is important because it estimates the effects of possible actions, which can guide which actions we choose to take.
The concept is a general one and includes predicting the response of a cell to a drug, as well as
predicting how a given patient is likely to respond to alternative surgical procedures, for example.

17. What is the Big Data Problem on which the CCD is Primarily Focused?

18. Number of variables (nodes)
The Number of Causal Model Structures as a Function of the Number of Measured Variables*
Number of variables (nodes)
Number of Causal Model Structures 1 2 3
* Assumes there are no latent variables and no directed cycles.

19. Number of variables (nodes)
The Number of Causal Model Structures as a Function of the Number of Measured Variables*
Number of variables (nodes)
Number of Causal Model Structures 1 2 3 25 4
543
* Assumes there are no latent variables and no directed cycles.

20. Number of variables (nodes)
The Number of Causal Model Structures as a Function of the Number of Measured Variables*
Number of variables (nodes)
Number of Causal Model Structures 1 2 3 25 4
543 5
29,281 6
3,781,503 7
1.1 x 109 8
7.8 x 1011 9
1.2 x 1015 10
4.2 x 1018
* Assumes there are no latent variables and no directed cycles.

21. Our Main Big Data Problem
Analyze biomedical datasets containing a large number of variables in order to generate plausible hypotheses of the causal relationships that hold among those variables
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

22. An Example Algorithm for Causal Discovery with Many Variables: FGES
GES: A popular CBN learning algorithm that uses greedy search and Bayesian scoring*
We developed a fast version of GES, called FGES
Optimized the single processor version of GES
Parallelized GES
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.
* Chickering DM. Optimal structure identification with greedy search. Journal of Machine Learning Research 3 (2002)

23. Evaluation of FGES Generated 10 random CBNs
30,000 nodes and 60,000 edges
Continuous-variables with linear relationships and Gaussian noise
Sampled each CBN to generate 1,000 cases
Provided those cases to FGES and measured its ability to learn the data-generating CBN
Average
Directed Arc
Precision
Average Directed Arc
Recall
# Processors
Average Learning
Time
99%
84%
128
2.3 minutes
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.
For more information:
Ramsey J, Glymour C. A Million Variables and More: The Fast Greedy Search (FGS) Algorithm for Learning High Dimensional Graphical Causal Models (to appear).

24. Another Example of an Algorithm for Causal Discovery with Many Variables: GFCI
FGES assumes there are no latent confounders, that is, there are no latent variables that cause two or more measured variables
Biomedical data often contain latent confounders
GFCI* allows for the possibility of latent confounders
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.
Ogarrio JM, Spirtes P, Ramsey J (2016). A hybrid causal search algorithm for latent variable models.
JMLR Workshop and Conference Proceedings, 52,

25. Evaluation of GFCI Generated more than 100 random CBNs
1,000 nodes and 2,000 edges
Continuous variables with linear Gaussian relationships
Sampled each CBN to generate 2,000 cases
Provided cases to GFCI and measured its performance
% Latent Nodes
Average Directed Arc
Precision
Recall
# Processors
Average Learning
Time 5%
92%
93% 1
15 seconds
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.
For more information:
Ogarrio JM, Spirtes P, Ramsey J (2016). A hybrid causal search algorithm for latent variable models. JMLR Workshop and Conference Proceedings, 52,

26. Ongoing Algorithm Work Includes …
Modeling non-linear relationships
Modeling causal feedback
Handling a mixture of continuous and discrete variables
Outputting uncertainty in edge relationships
Learning the causal relationships among latent variables
A group of investigators in the School of Medicine at the University of Pittsburgh will be submitting an application on the topic of modeling and discovery of causal networks from big biomedical data.
The primary aims will be to advance the representation, discovery, and uses of causal network models when applied to very large biomedical datasets.

27. Summary Causal discovery is central to biomedical science
The variety, richness, and quantity of biomedical data are increasing rapidly
The CCD is providing software now for analyzing big biomedical data to discover causal relationships
Causal discovery algorithms with additional capabilities will soon be available as well
A group of investigators in the School of Medicine at the University of Pittsburgh will be submitting an application on the topic of modeling and discovery of causal networks from big biomedical data.
The primary aims will be to advance the representation, discovery, and uses of causal network models when applied to very large biomedical datasets.

28. Acknowledgements
Thanks to the members of the Algorithms Group of the Center for Causal Discovery for their contributions to the activities described in this talk.
The Center for Causal Discovery is supported by grant U54HG008540 awarded by the National Human Genome Research Institute through funds provided by the trans-NIH Big Data to Knowledge (BD2K) initiative ( content of this presentation is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

29. CCD software is available at:
Thank you
CCD software is available at:
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

30. NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

31. Extra Slides
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

32. Association Versus Causation
Represents statistical relationships
Predicts outcomes from passive observations
Example uses: classification and regression
Causation:
Represents mechanisms
Predicts outcomes of active intervention
Example uses: decision making and planning
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

33. Example Association Causation Smoking – lung cancer – coughing
Both smoking and coughing predict lung cancer
Causation
Smoking  lung cancer  coughing
Smoking influences lung cancer
Coughing does not influence lung cancer
NIH recently announced a funding opportunity for developing methods that help derive biomedical knowledge from big data.
Six to eight Centers of Excellence will be supported for up to 4 years starting next year.

34. Recent Examples of the Use of Graphical Causal Discovery Methods
Anticipation-related brain connectivity in bipolar and unipolar depression: A graph theory approach
Anna Manelis, Jorge R. C. Almeida, Richelle Stiffler,1 Jeanette C. Lockovich, Haris A. Aslam, Mary L. Phillips.
Brain 139 (2016)
Dobryakova, E., Costa, S. L., Wylie, G. R., DeLuca, J., & Genova, H. M. (2016). Altered effective connectivity during a processing speed task in individuals with multiple sclerosis. Journal of the International Neuropsychological Society: JINS, 22(2),
Otsuka, J. (2016). Discovering phenotypic causal structure from nonexperimental data. Journal of evolutionary biology, 29(6),
Attur, M., Statnikov, A., Samuels, J., Li, Z., Alekseyenko, A. V., Greenberg, J. D., et al. (2015). Plasma levels of interleukin-1 receptor antagonist (IL1Ra) predict radiographic progression of symptomatic knee osteoarthritis. Osteoarthritis and Cartilage, 23(11),