lcome to Rocky 1
Jan 22, 2016
Welcome to Rocky 1
“Go to the Mountains and Get their Good Tidings”
Inspirations:
Adrenaline
Beauty
Life
– John Muir
Inspiration for a Revolution!
Science is in the midst of a tremendous explosion of knowledge regarding of life
Exponentially growing knowledge challenges humanity’s ability to integrate and appreciate it
Our era cries out for big ideas
Aristotle 384-322 BC
Linnaeus1707-1778
Darwin1809-1882
Mendel1822-1884
Franklin1920-1958
A timeline and some great minds of biology
Why Rosalind Franklin?
Women’s enormous contributions to the study of life have often been downplayed
Before she died of cancer at age 37, she produced the first X-ray crystal structure of DNA
Watson and Crick were shown this image shortly before they produced their double-helix model
Data drives modeling…
The challenge of exponentially growing knowledge
Numbers are articles from a given year.
Fits an exponential curve with a 4.32% growth rateR2 = 0.998
Life is deeply connected:High order interactions dominate
Unsuspected connections in the last 3 years:– Uber-oncogene P53 plays an important role in aging– Expression array studies of remodeling cardiac tissue
after heart failure implicate role for genes well studied in pregnancy and embryological development
– Panadol, a drug developed for cardiovascular illness turn out to be very important in the treatment of depression.
Gene-gene (or protein-protein) interactions are not pairwise, but very high order (often >10)
Towards The Biological Knowledge-base
Inferential potential of a unified knowledge-base transcends human ability– Even heroic bioscientists can’t keep up with flood of
information as disciplinary boundaries break down.
Computational integration efforts– SOAP, GRID and especially the Semantic Web
Beyond integration– Knowledge dissemination: timing and comprehensibility– Making a compelling story from disparate bits of evidence
Biognostic Machines:An AI Vision for Bioinformatics
From the Greek(life) and(knowing)
The integration of humanity’s knowledge of life in a computational system that can interact with bioscientists as a knowledgeable colleague– Keeps up with the literature– Can provide explanations and evidence for its statements– Transcends disciplinary and terminological boundaries
AI to the rescue?
A bit of AI
Cognitive systems are driven by “goals”
Experience, knowledge, memory, practice, learning, etc. inform both perception and action
Sense perception provides incomplete and error-prone information about the world
Action is organized and controlled to achieve goals (perhaps opportunistically)
Mind is many distinct processes working together
Biognostic AI
Goals:– Improve human health, diagnose and treat disease– Pharmaceuticals: their design and improvement– Causal generalizations, understanding
Experience (knowledge, memory, etc.):– Up-to-date fact/knowledge-base, from textbooks, domain
experts, journal articles, other databases– Library of physical, statistical & logical models and
classes of models– Sets of models & parameters for particular applications
Biognostic Sensation
Sensation is the use of pattern recognition (statistics) and knowledge to recognize opportunities for achieving goals via perceptions
Biognostic Perceptions:– The biomedical literature (via information extraction) – Databases: GO/A, GenBank, expression databases, etc.– Sense vocabulary: GO, UMLS, NCI common data
elements, ESV vocabulary, MAGE-ML, etc.– Instruments? MS, NMR, etc. (or better from databases?)
Biognostic Actions & Abilities
Extract information from the literature
Select models, fit parameters from data– Learning, optimization, model competition
Simulation / Prediction– Application of models to unobserved circumstances
Creation of new classifications or categorizations
Communicating– Explain, justify, answer questions, visualize/diagram
Design experiments & monitor or control instruments?
Vision versus Speculation
Vision is necessary for engineering the tools to achieve it. Speculation is ungrounded and a distraction from doing the work
Sometimes hard to tell the difference…
Biognostic machines may be vision, since – Many pieces starting to fall into place: Ontology,
information extraction, semantic web, etc. etc.– We are not alone: Paul Allen’s Project Halo
Guide to next few days
Purpose of the meeting is to build community– Get to know each other’s names, work, institutions– Find common interests and potential collaborations– Let your hair down, have big ideas, have fun!
Afternoons are part of the program:– Informal interactions are just as important as talks– Good skiers: find Elvis & Marylin shrines (Back of Bell)– Novices: create a small group (4-8) for a joint lesson. – Enjoy the town: it’s easily walkable.
Thanks!
International Society for Computational Biology Stephanie Hagstrom
CU Center for Computational Biology Stephen Billups
IBM (for dinner!), Kirk Jordan, Alex Zekulin, and the rest…
Apple and the other sponsors