Makefile lab meeting

12 noon, Wednesday 4/4, 2007, 425 Hearst Mining Building (see LabMeetings)

The topic of this meeting was the use of automated build tools for bioinformatics pipelines. (See, for example, this intro on nodalpoint.org, or this article by Chris Lee.)

Conclusions

The conclusions of the meeting are summarized on the MakeComparison page.

Miscellaneous notes

George recommended Rose-DB as a Perl database interface.

Agenda

Proposed assignments in parentheses.

• Why make?
  • Bioinformatics-in-the-large by Lee et al (PDF) (Ian)
  • Make use cases: Rfam analyses, fly zipcodes, fly RNA scan... (Ian)
  • How make should work: BioMake and Skam (Ian, with backup from Chris)
  • What's wrong with GNU make: MakefileManifesto (Andrew)
• Parallel processing
  • Salient features of Sun grid engine (with examples) (Andrew)
  • Parallel and/or improved make clones, including omake, distmake, qmake (Avinash)
• Roll-your-own make replacements
  • Perl (BDGP SapsModules; makepp) and Python (SCons) systems (Rob, Ian; others?)
  • RDBMS extensions: BioPostgres; PostMake and PostModel (George, Andrew, Ian)
  • ApacheAnt, other make alternatives incl. Jam, Cook (Yuri)
  • RubyMake (George? Ian? Mitch? anyone...)
• Utopias
  • Google's map reduce and file system (Ian)
  • Functional programming languages: ErlangLanguage, TermiteScheme (Mitch)
  • The GUI approach: BioinformaticsWorkflows, Taverna etc. (Yuri)
• Realities...?
  • SummerOfCode

A few practical issues to think about:

• What's real (i.e. implemented)?
• What's re-useable?
• What fancy systems could be real, with a little work?
• What could you do with those fancy systems that you couldn't do before?

-- MadeGuys - 21 Mar 2007