Revision as of 09:20, 27 October 2010 by AlejandroHD (Talk | contribs)

Modeling Notebook

Welcome to the notebook of the modeling project. This notebook includes the work in four different tracks: miBEAT, miRockdown, miBS Designer and the model training set.



  • Meeting with Nao
    • Discuss about last year's modeling approach
    • Problems to be avoided



  • Discussion about first model idea
The first model concept


  • Discussion about programming languages
    • Advantages of R, Perl and MatLab
    • Decision to use Perl and Matlab
    • Request for Matlab license


First week

  • Activation of MatLab-Accounts
  • Start working on several tutorials

Second week

  • Meeting with Dr. König
    • Advantages and disadvantages of different modeling approaches
    • Possibility to use Neuronal Networks and Support Vector Machines

Third week

  • Activation of MatLab fuzzy logic toolbox

Fourth week

  • Meeting with Tim and Nao
    • Options to optimize a fuzzy logic model
  • Start of subproject miBSdesigner


  • Telemeeting with Stephen Krämer
    • Advantages and limitations of fuzzy logic



  • Project miBEAT starts
    • decision to create an interactive GUI on the wiki
    • creation of test GUI using javascript
  • Activation of MatLab neural net toolbox
    • Start of Neural Network model


  • Presentation of miBEAT project-concept in official team meeting
  • miBS designer is finished
  • mUTING subproject is started


  • 6th Workshop on Computation of Biochemical Pathways and Genetic Networks
    • Parameter estimation in fuzzy/noisy systems
    • Global sensitivity analysis
    • New approaches to spatial modeling
    • Computation of microRNA regulatory network
  • Discussion with Professor Sanghamitra Bandyopadhyay from the Machine Intelligence Unit of the Indian Statistical Institute in Kolkata
    • Advise to use a Rough set modeling approach


  • Project miRockdown Starts
    • Development of an modified miBS designer to be part of miBEAT
    • Development of miRockdown concept
    • Decision on the use of the modified targetscan_50_context_scores–Algorithm for binding site evaluation


  • Both models (neural network and fuzzy logic) are ready and waiting for data
  • Planning for normalization strategies
  • Designs of Binding Sites to be tested in wet lab