Pipe+Presentation

intro

  • describe the niche we're trying to fill
  • stay away from any explanation/justification of technology use, java, swing, etc.
  • be oober conceptual, we can entertain techie questions at the end or at a different meeting altogether.

basic concepts

  • ad hoc queries
  • workbench
  • modules

use simple example

  • get 25 sequences from genbank
  • align to ref sequence
  • take top 5 hits
  • perform customoperation X

--old way--- go to browser -> genbank and use "File Save As:" 25 times--- or write database access code specific to db--- run blast on it, --- write a parser to parse the output for your custom operation's input--new way--- identify accession numbers, give it to genbank module--- connect to blast module --- connect to custom module--- only code you wrote was for your own custom module!

  • change of parameters, different sequences, different database, different alignment tool

--old way--- go back and do "File Save As:" another 25 times.--- rewrite all db access code--- rewrite all parsing code.--new way --- rewrite nothing! just download new modules that came with the new tools.

advanced concepts

  • general purpose modules

-- aggregators/splitters/iterators

  • saving results (i.e. saving state)
  • sharing analysis protocols

successful models

alternatives and why they dont fill our niche:

  • www.scitegic.com - chem informatics plus it costs money
  • biomoby - web service
  • biopipe - focussed on batch processing
  • biojava - unstable
  • jpipe - focussed on farming processes

resources

  • lacroix's book
  • national graduate fellowship
  • www.bsml.org
  • other previous attempts to create a biological data representation language

things we need to get done

  • conceptual model
  • backend
  • modules
  • gui

goals

-- the soon to be ncbi for bioinformatics

  • allow scientists to do bioinformatics
  • save headache
Posted: November 19, 2007 | | comments