1. A highly interdisciplinary open road through Computational Science
Bradly Alicea

2. Outline for Talk DevoWorm/OpenWorm (Developmental Biology, Evolution,
and Digital Biology)
Personal Research
Context
MIND and Cellular Reprogramming Labs (Cognitive Systems/Systems
Biology)
Vision for a Research Program

3. Personal Research Context
Cognitive Systems and
Virtual Reality
Systems Biology of
Cellular Reprogramming
DevoWorm
(Developmental Biology, Evolution,
and Artificial Life)

4. Cognitive Systems/Systems Biology
MIND Lab (Michigan State University)
Virtual Reality
(Headset and CAVE Environment)
Augmented Cognition
(attention, spatial cognition, and motor learning)
Cognitive Neuroscience
(among expert video game players)
1.1

5. Virtual Environments Enable us to Manipulate Sensory Information and Cognitive Representations
Review article that describes the use of virtual reality in understanding the brain.
* focus is on the use of virtual reality as stimulus generation in neuroscience research (human and non-human animals).
* dovetails with other work on physiological measurements (fNIR, electromyography) in virtual environments.
Bohil, C., Alicea, B., and Biocca, F. Nature Reviews Neuroscience, 12, (2011). Invited Review and Cover Art Inspiration.
1.2

6. (program, representation)
Virtual Environments Enable us to Manipulate Sensory Information and Cognitive Representations
Virtual Environment
Nervous System
Content
(program, representation)
INITIAL
EXPOSURE
Stimuli
(multisensory
features)
1.3

7. (program, representation)
Virtual Environments Enable us to Manipulate Sensory Information and Cognitive Representations
Virtual Environment
Nervous System
Content
(program, representation)
INITIAL
EXPOSURE
Sensory System
(dynamic range)
Stimuli
(multisensory
features)
FEEDBACK
Brain Network
(connectivity,
representation)
This gives us the capacity to run dynamic, real-time experiments involving perception and cognition with relatively inexpensive technology in minimal space.
1.4

8. Interest in Biological Systems at Multiple Scales
Social
Behavior
Cellular
Molecular
1.5

9. Interest in Biological Systems at Multiple Scales
Social
Cognitive Systems
Behavior
Cellular
Systems Biology
Molecular
1.6

10. Cognitive Systems/Systems Biology
Cellular Reprogramming Lab (Michigan State University)
mRNA Regulation in
Transcriptome and Translatome
Cellular Reprogramming
Comparative Gene
Expression
1.7

11. Cellular Reprogramming Laboratory (Michigan State University)
Cellular Diversity, Comparative Bioinformatics, Models for Comparing Different Types of Reprogramming
Example of work on cellular reprogramming:
1.8

12. Cellular Reprogramming Laboratory (Michigan State University)
Cellular Diversity, Comparative Bioinformatics, Models for Comparing Different Types of Reprogramming
Example of work on cellular reprogramming:
* introduce transcription factors into fibroblast cells to
induce cells of a different type (skeletal muscle – iSM,
neuronal – iN, and pluripotent – iPS).
1.9

13. Cellular Reprogramming Laboratory (Michigan State University)
Cellular Diversity, Comparative Bioinformatics, Models for Comparing Different Types of Reprogramming
Example of work on cellular reprogramming:
* introduce transcription factors into fibroblast cells to
induce cells of a different type (skeletal muscle – iSM,
neuronal – iN, and pluripotent – iPS).
* reprogramming occurs at a certain efficiency (number
of induced cells per all cells infected).
* is this efficiency variable across cells from different
source tissues, genotypic backgrounds, and species?
1.10

14. Induced Neuronal Cells (iNCs)
Induced Neurons, Different Genotypes and Different Human Sources
Induced Neuronal Cells (iNCs)
Induced Neurons, Single Genotype, Different Tissues in Same Mouse
Alicea, B., Murthy, S., Keaton, S.A., Cobbett, P., Cibelli, J.B., and Suhr, S.T.   Stem Cells and Development, 22(19), (2013).
1.11

15. Induced Skeletal Muscle Cells (iSMCs)
Induced Muscle Fibers, Different Genotypes and Different Human Sources
Induced Muscle Fibers, Single Genotype, Different Tissues (e.g. KIdney) in Same Mouse
Alicea, B., Murthy, S., Keaton, S.A., Cobbett, P., Cibelli, J.B., and Suhr, S.T.   Stem Cells and Development, 22(19), (2013).
1.12

16. Normalized Reprogramming Efficiency
Normalized Reprogramming Efficiency(reprogrammed cells/all cell nuclei)
Human Cells: variable genetic background, same tissue (more uniform).
Mouse Cells: same genetic background, different tissues (less uniform).
Take-home messages:
* diversity is non-uniform across cell lines.
* non-uniformity = different tissues of origin > genetic background.
Normalized Reprogramming Efficiency
1.13
Individual Cell Lines

17. Modeling of Non-uniform Response
LEFT: exponential statistical models (Poisson exact test, jackknife resampling, 10,000 replicates).
Converted Cell Area (um2)
1.14

18. Modeling of Non-uniform Response
LEFT: exponential statistical models (Poisson exact test, jackknife resampling, 10,000 replicates).
RIGHT: can also be modeled as a conversion rate per unit time (variable across replicates).
Converted Cell Area (um2)
Conversion Rate/Hour
1.15

19. Model: accumulation of individuals with induced state per population over time
H0: “uniform” traffic flow (jam) – normal distribution of cars per unit time.
Suggests cell-to-cell homogeneity in length of reprogramming process.
H1: “non-uniform” traffic flow – non-normal distribution of cars per unit time.
Suggests cell-to-cell variability in length of reprogramming process.
1.16

20. The “transcriptional state” of an iPS cell
DATA SOURCES: multiple whole-genome microarrays for each human cell type
(secondary data).
Alicea, B. and Cibelli, J.B., Principles of Cloning, 2nd edition, Chapter 37 (2013).
Q: What are the differences between pluripotent stem lines (define the parameters of the stem-like state)?
A: Do a comparative analysis using selected secondary data.
Use criteria (threshold, distance metric) to determine genes that underpin iPS, ES, and 8 cell (totipotent) cellular states.
iPS-ES
iPS-8 cell
ES-8 cell
8-cell embryo is totipotent, serves as the baseline for a state of stemness.
1.17

21. Establish quantitative differences for pairwise comparisons of gene expression profiles between cellular states:
* iPS “closer” to 8 cell than ES, iPS shows a similar number of differences compared to ES cell state.
GO analysis: about 50% of pairwise comparison differences are ribosomal proteins, slightly less for ES-8 cell comparison.
Ingenuity (IPA) analysis: iPS-8 cell/iPS-ES and ES-8 cell comparisons associated with different gene networks.
Among human
cell lines
Around 1% of all genes = ribosomal proteins. Elevated proportion in our comparisons. Important supportive functions?
1.18

22. The “transcriptional state” of an iPS cell
Alicea, B. and Cibelli, J.B., Principles of Cloning, 2nd edition, Chapter 37 (2013).
Mutual Information (MI) within and between cell lines:
Mutual Information: shared information (or variability) between two ensembles.
* analysis conducted in MATLAB (pairwise comparisons between microarrays and between cell types).
1.19

23. The “transcriptional state” of an iPS cell
Alicea, B. and Cibelli, J.B., Principles of Cloning, 2nd edition, Chapter 37 (2013).
Mutual Information (MI) within and between cell lines:
Mutual Information: shared information (or variability) between two ensembles.
* analysis conducted in MATLAB (pairwise comparisons between microarrays and between cell types).
EXAMPLE: iPS, ES – the amount of information shared between the iPS and ES cell lines.
H(X) = iPS
H(Y) = ES
H(X1,X2,…Xn) = within iPS
H(X,Y) = iPS and ES
1.20

24. Single, exact measurements
WITHIN
CELL LINES
Pairwise MI analysis tells us:
* 8-cell lines seem to share less information with iPS, ES lines (totipotency vs. pluripotency).
* less pronounced differences when comparing between cell lines.
Core regulatory machinery: does not change between cell types.
* changes are not due to pluripotency per se.
Single, exact measurements
(no error bars)
BETWEEN
CELL LINES
1.21

25. Interest in Biological Systems at Multiple Scales
Social
Behavior
Phenotypes and Genotypes
Cellular
Systems Biology
Molecular
1.22

26. DevoWorm/OpenWorm COURTESY: OpenWorm Browser, Christian Grove, Caltech
Models rendered in Blender
EXAMPLE: approximation with phenotypic modeling
2.1

27. Merges my interests in virtual reality with systems biology.
OpenWorm Foundation: umbrella including project that apply computational modeling and data analysis to simulating ~1000 cell organism at the cellular level.
PUBLICATION: OpenWorm: an open-science approach to modeling Caenorhabditis elegans. Frontiers in Computational Neuroscience, doi: /fncom
Simulating a simple
Metazoan lifeform
2.2

28. Geppetto Project Simulating a simple Metazoan lifeform
Multi-scale simulator of C. elegans nervous system
Simulating a simple
Metazoan lifeform
Sybernetic Project
Biomechanics of movement in C. elegans
WormSim Project
Nervous system explorer for C. elegans
2.3

29. Merges my interests in virtual reality with systems biology.
Currently a Foundation, a need to model the C. elegans embryo in a manner that matches the adult.
Many opportunities for student research, from low-hanging fruit to deeply thought-out projects:
* wet-lab experimentation + imaging.
* bioinformatics and computational modeling.
* development of new computational methods.
Collaborative opportunities with distributed, collaborative teams, in additional to other biological and computational labs across the globe.
Simulating a simple
Metazoan lifeform
NeuroML +
WormBrowser
2.4

30. DevoWorm (http://devoworm.weebly.com)
Developmental
Dynamics (models
extracted from data)
Cybernetics and
Digital Morphogenesis
(Morphozoic)
Loose international (Canada, USA, Japan) research
collaboration of biologists, computer scientists, and
science advocates.
2.5

31. analysis (embryogenesis)
Comparative
Developmental
Biology
Secondary data
analysis (embryogenesis)
2.6

32. Network Analysis
Interactions
between cells in
embryo
2.7

33. Developmental Plasticity (larval arrest): mutant strains
2.8

34. Experimental Evolution: mutant strains
2.9

35. Simulation (Cellular Automata + Artificial Neural Networks)
2.10

36. High-resolution imaging of Embryos
IMAGE: Nature, 442(7103), (2006).
2.11

37. DevoWorm ties together data, models, and techniques from across several disciplines
* developmental and evolutionary biology, genetics, computer science, and physics.
PREPRINT: DevoWorm. bioRxiv, doi: / (2014).
PUBLICATION: Quantifying Mosaic Development. Biology, 5(3), 33 (2016).
2.12

38. Development of Caenorhabditis elegans
Each cell division contributes to a
deterministic lineage tree (invariant
between embryos)
Adapted from:
bastiani/3230/DB%20Lecture/Lectures/b12Worm.html
From embryo
2.13

39. Development of Caenorhabditis elegans
Adult organism is eutelic
(always 959 cells in hermaphrodite,
1024 cells in male)
Courtesy:
ResearchF/elegans.html
To adult
2.14

40. Development of Caenorhabditis elegans
200 minutes
250 minutes
300 minutes
350 minutes
400 minutes
Differentiation and growth processes for different tissues can be captured statistically
2.15

41. Comparative Development
Axolotl
(dataset from Richard Gordon)
Differentiation tree:
In doi: /biology
Ciona Intestinalis
(dataset from ANISEED)
Differentiation tree:
doi: /m9.figshare
Caenorhabditis elegans
(dataset from Zhirong Bao)
Differentiation tree:
doi: /m9.figshare
COURTESY:
2.16

42. Comparative Development
Drosophila melanogaster
Dataset (imaginal disc only) from
Wolff, 1993
Differentiation tree:
TBD
Mouse
Dataset from doi: / nmeth.3690
Differentiation tree:
TBD
COURTESY:
2.17

43. Empirical Science: Experimental Evolution of Fecundity
Genotype-dependent increase due to selection for fecundity over short evolutionary timescales.
Wild-type (N2) versus genetic mutants of known function.
PUBLICATION: Royal Society Open Science, 3,
2.18

44. PREPRINT: bioRxiv, doi:10.1101/045609
Empirical Science: Modeling of Developmental Plasticity and Stress Response
Components of change in fecundity due to starvation-induced larval (L1) arrest for wildtype genotype (N2).
Data available for both wildtype and genetic mutants of known function.
PREPRINT: bioRxiv, doi: /045609
2.19

45. DevoWorm: Future Directions
Cybernetic
Embryo
Cellular-level
Physics Models
(CompuCell3D)
Computational
Models of Self-
Organization
Data Science of
High-resolution
Imaging
Secondary Data
Integration
Picture courtesy: Annie Lam, Chin-Sang
Lab, Queens University, Canada
2.20

46. Research Program Vision
Current Resources:
Collaborative expertise (multiple areas).
* some promising avenues for future projects (complex systems and artificial life, systems biology, data science).
External
Collaborators
Department, Program,
University
3.1

47. Research Program Vision
Current Resources:
Collaborative expertise (multiple areas).
* some promising avenues for future projects (complex systems and artificial life, systems biology, data science).
Network of methods and learning opportunities.
* infrastructure of educational, research opportunities (e.g. OpenWorm).
3.2

48. Goals For the Next Several Years
Build upon external collaborative opportunities with other academic labs and OpenWorm.
* other interdisciplinary initiatives.
Develop funding streams.
* multi-institution grants.
Focus on further development of most promising techniques and experimental approaches.
“To have diverse research experiences
and promote interdisciplinary collaboration”
3.3

49. Approach to Interdisciplinarity
COMPUTATION
Research Program Theme: Exploration of the Biology – Theory – Computational interface
Student researchers interested in one aspect (e.g. biology) can gain exposure to others.
Gain unique experience in theory-building, developing quantitative models, informatics skills.
THEORY
BIOLOGY
“To explore intellectual frontiers while learning how to produce cutting-edge research”
3.4

50. Goals For Advancing Interdisciplinarity
Lab design: wet lab and computational lab components.
* unique opportunity to do computational biology and cognitive neurobiology at multiple scales.
Flexibility: emphasis on specialties would be contingent upon funding, current opportunities.
* example: microscopy data can be acquired via collaboration, while analysis could be in-house.
“To overcome unfair characterizations and
become methodologically versatile”
3.5

51. Acknowledgements
Dr. Frank Biocca
Dr. Corey Bohil
Dr. Rene Weber

52. Acknowledgements Dr. Frank Biocca Dr. Corey Bohil Dr. Jose Cibelli
Rebecca Androwski
Dr. Frank Biocca
Dr. Corey Bohil
Dr. Jose Cibelli
Sarah Keaton
Shashanka Murthy
Dr. Steven Suhr
Dr. Rene Weber

53. Acknowledgements Rebecca Androwski Dr. Frank Biocca Dr. Corey Bohil
Dr. Jose Cibelli
Dr. Richard Gordon
Sarah Keaton
Dr. Stephen Larson
Shashanka Murthy
Dr. Tom Portegys
Dr. Steven Suhr
Dr. Rene Weber

54. Thank You for Your Attention!
COURTESY: White noise mask and Figure 1, Journal of Neurophysiology,
85(2), (2001).