Tracing the Origins of Data and Ideas - Provenance Visualization for Biomedical Discovery

Provenance Visualization for Biomedical Discovery

HARVARD MEDICAL SCHOOL DEPARTMENT OF BIOMEDICAL INFORMATICS

Nils Gehlenborg・http://gehlenborglab.org・@nils_gehlenborg

Tracing the Origins of Data and Ideas

http://gehlenborglab.org

Nature asked 1,576 researchers if there is a reproducibility crisis in science.

M Baker, Nature 533, 452-454, 2016

0% 100%

No crisis (3%)

Don’t know (7%)

Slight crisis (38%)

M Baker, Nature 533, 452-454, 2016

Significant crisis (52%)

Nature asked 1,576 researchers if there is a reproducibility crisis in science.

M Baker, Nature 533, 452-454, 2016

M Baker, Nature 533, 452-454, 2016

Intentional?

Inability to capture everything?

Inability to communicate everything?

M Baker, Nature 533, 452-454, 2016

Intentional?

Inability to capture everything?Inability to communicate everything?

SOCIAL ISSUE

TECHNICAL ISSUES

PROVENANCE

The term, provenance, has been used in a variety of ways to describe different types of origins and histories.

ED Ragan et al., IEEE Transactions on Visualization and Computer Graphics 22, 31 – 40, 2015

Types of Provenance Information

ED Ragan et al., IEEE Transactions on Visualization and Computer Graphics 22, 31 – 40, 2015

Data Visualization

Interaction

Insight

Rationale

COMPUTATIONAL WORKFLOWS

EXPLORATORY VISUALIZATION


Data Visualization

Interaction

Insight

Rationale

COMPUTATIONAL WORKFLOWS

EXPLORATORY VISUALIZATION


ROLE OF VISUALIZATION

PUBLICATION

experiment

DATA

INSIGHT HYPOTHESIS

interpretation

hypothesis generation

PUBLICATION

experiment

DATA

INSIGHT HYPOTHESIS

interpretation


PRESENTATION“Storytelling”

experiment

DATA

INSIGHT HYPOTHESIS

interpretation


EXPLORATION“Pattern Discovery”

experiment

DATA

INSIGHT HYPOTHESIS

interpretation

HYPOTHESIS


EXPLORATION

HYPOTHESIS-DRIVEN DISCOVERY

“Pattern Discovery”

experiment

DATA

INSIGHT HYPOTHESIS

interpretation

DATA


EXPLORATION

DATA-DRIVEN DISCOVERY


The Cancer Genome Atlas

10,000+

genomes + clinical data + molecular data

CANCER SUBTYPES

mRNA expression microRNA expression

DNA methylation

protein expression

copy number variants mutation calls

clinical parameters

mRNA expression microRNA expressionDNA methylation

protein expressioncopy number variants mutation calls

clinical parameters

C4C3C2C1

mRNA expression - clustering

C4C3C2C1

mRNA expression

copy number variants

DEL NORMAL AMP

- clustering

- gene X

DEL NORMAL AMP

C4C3C2C1

mRNA expression


mutation calls

WILDTYPEMUT

- clustering

- gene X

- gene Y

DEL NORMAL AMP

C4C3C2C1

mRNA expression


mutation calls

WILDTYPEMUT

- clustering

- gene X

- gene Y

DEL NORMAL AMP

C4C3C2C1

mRNA expression


mutation calls

WILDTYPEMUT

- clustering

- gene X

- gene Y

PROBLEM 1

Visualize overlap of patient sets across two or more stratifications.

PROBLEM 2

Visualize characteristics of patient sets within a stratification of interest.

A Lex, M Streit, H-J Schulz, C Partl, D Schmalstieg, PJ Park, N Gehlenborg, Comput Graph Forum, 2012 M Streit, A Lex, S Gratzl, C Partl, D Schmalstieg, H Pfister, PJ Park, N Gehlenborg, Nat Methods, 2014

Divide & Conquer Visualization: StratomeX

PROBLEM 1

Visualize overlap of patient sets across two or more stratifications.

PROBLEM 2

Visualize characteristics of patient sets within a stratification of interest.

PROBLEM 3

Identify relevant stratifications, pathways, and clinical variables.

Is there a mutation that overlaps with this mRNA cluster?

Is there a CNV that affects survival?

Is there a pathway that is enriched in this cluster?

Is there a mutually exclusive mutation?

Query

Stratifications Clinical Params

Pathways

GUIDED EXPLORATION

M Streit, A Lex, S Gratzl, C Partl, D Schmalstieg, H Pfister, PJ Park, N Gehlenborg, Nat Methods, 2014

Query

Rank

VisualizeStratifications

Clinical Params Pathways

GUIDED EXPLORATION

M Streit, A Lex, S Gratzl, C Partl, D Schmalstieg, H Pfister, PJ Park, N Gehlenborg, Nat Methods, 2014

AND NOW WHAT?


experiment

DATA

INSIGHT HYPOTHESIS

interpretation

DATA


EXPLORATION“Pattern Discovery”


PUBLICATION

experiment

DATA

INSIGHT HYPOTHESIS

interpretation

DATA


EXPLORATION

PRESENTATION“Storytelling”



DATA-DRIVEN COMMUNICATION

finding figure/videoAuthoringExploration Presentation




Current Model



finding figure/videoAuthoringExploration PresentationWhat we show.




What we tell.




What we did.



track provenance

annotate observations

make sense of observations

tell the story



Capture

Label

Understand

Explain

track provenance



tell the story



Capture

Label

Understand

Explain

track provenance



tell the story

C

L

U

E

CLUE

vistories

Authoring

Exploration Presentation



CLUE Model

Exploration

Authoring

Presentation

Exploration

Authoring

Presentation

Exploration

Authoring

Presentation

Exploration

Authoring

Presentation

Exploration

Authoring

Presentation

Exploration

Authoring

Presentation

Exploration

Authoring

Presentation

Exploration

Authoring

Presentation

VISTORY = visualization + story + history

VISTORY = visualization + story + history

Do collaborative data analysis.

Use during peer-review.

Publish with a paper.

Embed in a presentation.







http://vistories.orgDemos and prototypes built with

SAMUEL GRATZL JOHANNES KEPLER UNIVERSITY LINZ

ALEXANDER LEX UNIVERSITY OF UTAH

MARC STREIT JOHANNES KEPLER UNIVERSITY LINZ

HOLGER STITZ JOHANNES KEPLER UNIVERSITY LINZ

My lab is hiring postdocs!

HARVARD MEDICAL SCHOOL DEPARTMENT OF BIOMEDICAL INFORMATICS

See http://gehlenborglab.org or http://dbmi.med.harvard.edu for details.

Data visualization, analysis, and management for: • exploration tools for data repositories • provenance graphs • genomic structural variants • dynamics of the 3D genome • cancer subtypes in patient cohorts

Tracing the Origins of Data and Ideas - Provenance Visualization for Biomedical Discovery

Science