Analysis, Biomedicine, Collaboration, and Determinism Challenges and Guidance: Wish List for Biopharmaceuticals on the Interface of Computing and Statistics

Abstract

I have personally witnessed processing advance from desk calculators and mainframes, through timesharing and PCs, to supercomputers and cloud computing. I have also witnessed resources grow from too little data into almost too much data, and from theory dominating data into data beginning to dominate theory while needing new theory. Finally, I have witnessed problems advance from simple in a lone discipline into becoming almost too complex in multiple disciplines, as well as approaches evolve from analysis driving solutions into solutions by data mining beginning to drive the analysis itself. How we do all of this has transitioned from competition overcoming collaboration into collaboration starting to overcome competition, as well as what is done being more important than how it is done has transitioned into how it is done becoming as important as what is done. In addition, what or how we do it being more important than what or how we should actually do it has shifted into what or how we should do it becoming just as important as what or how we do it, if not more so. Although we have come a long way in both our methodology and technology, are they sufficient for our current or future complex and multidisciplinary problems with their massive databases? Since the apparent answer is not a resounding yes, we are presented with tremendous challenges and opportunities. This personal perspective adapts my background and experience to be appropriate for biopharmaceuticals. In these times of exploding change, informed perspectives on what challenges should be explored with accompanying guidance may be even more valuable than the far more typical literature reviews in conferences and journals of what has already been accomplished without challenges or guidance. Would we believe that an architect who designs a skyscraper determines the skyscraper's exact exterior, interior and furnishings or only general characteristics? Why not increase dependability of conclusions in genetics and translational medicine by enriching genetic determinism with uncertainty? Uncertainty is our friend if exploited or potential enemy if ignored. Genes design proteins, but they cannot operationally determine all protein characteristics: they begin a long chain of complex events occurring many times via intricate feedbacks plus interactions which are not all determined. Genes influence proteins and diseases by just determining their probability distributions, not by determining them. From any sample of diseased people, we may more successfully infer gene probability distributions than genes themselves, and it poses an issue to resolve. My position is supported by 2–3 articles a week in ScienceDaily, 2011.

Key Words:

• Analysis
• Biomedicine
• Biopharmaceutical challenges
• Biopharmaceutical guidance
• Biopharmaceutical wish list
• Collaboration
• Computing and Statistics
• Data mining and statistics
• Determinism
• Informed future perspective
• Informed personal perspective

ACKNOWLEDGMENTS

I dedicate my perspective to the memory of Paul Silverman, who placed his vision for molecular biology into my mind, heart, and soul. Without it, our first process or systems model of cellular protein cycle and my seven-year advocacy of genetic uncertainty would not have existed. Paul had the insight to see genetic determinism decaying, courage to proclaim this widely, and perception to know he needed someone who knew both statistics and systems for his quest. This article is scheduled for publication closely following the seventh anniversary of his leaving our quest.

However, I could not have had the opportunity to meet and collaborate with Paul were it not for Robert Newcomb's inviting me to join him in co-founding the UCI Center for Statistical Consulting. Our center actively contributed not only to creating a Department of Statistics, but also to locating this department within the Bren School of Information and Computer Sciences—which forms a unique academic interface of computer science and statistics.

Notes

Arnold Goodman is the founder of Interface (1965), co-founder of “Interface Symposia” (1967), and founding editor of statistical analysis for Statistical Analysis and Data Mining Journal (2006).