Transcript Strategic Planning

R&D via Open Source
Software
Robert Gentleman
1
Overview
 R&D
 is expensive
 typically involves many false starts
 requires very flexible software infrastructure and access to
data/problems
 Open Source Software via Bioconductor is one way to achieve these
aims
 access to a broad community of users
 flexible software with good visualization and statistical routines built in
 active developer environment
 good data and code sharing opportunities
2
Then what
 once a methodology is viewed as being of general value efforts should
be made to commercialize it, make it fast, user friendly, etc
 moving too quickly into this area is expensive and if the methods don’t
pay off, then the investment is wasted
3
How
 most importantly common data structures need to be adopted
 they need not be fixed in stone, but we want everyone to be using
something similar, regardless of language (R, Java, C, Python)
 this means that we can easily translate data sets between software
packages and allows for shared development
 each different person, working on a different project knows that they will
be able to access and share data and hence we move towards
interoperability (R calls Java, C calls R, Java calls Python)
4
Tomorrow
 FHCRC, off Fairview Ave, the Arnold Building, 9:30~4:00
 we have to escort you in and out, so please be prompt
 Nov 7, at FHCRC we will have a one day scientific meeting at the
FHCRC, please contact me if you are interested in attending
5