Log of 2010-03-10

• 1. Peregrine
  1. obtained from ​https://gforge.nbic.nl/svn/data-mining/trunk/.to directory peregrine . (everything included inside )
  2. MAil question about the UI .
  3. TODO :
     1. search for UI in source code . RuN it .
     2. Clean eclipse projects from old downloads of pelegrine

1. Facebook for diseases

• When a patient comes into the practical of a clinical geneticist, typically an array of disease databases like OMIM is checked to classify their symptoms as a known syndrome. Diagnosis is then confirmed using genetic tests. However, in rare cases a unique set of symptomes is apparant in one individual or family which cannot be identified to be a known syndrome. Then the clinical geneticist has no choice but to search in literature or check with collegue geneticists if any case is known that appears to be similar. This is obviously hard and often the case is not solved and the patient not helped while there may be similar cases elsewhere on the world. However this search is greatly aggrevated by the fact that when dealing with new syndromes usually no standard terminology is used.
• Objectives:
  What is needed is a web-based database where clinicians can make 'mini-publications' describing their cases such that they can be shared and searched.
  The desired features:
  (1) Upload case descriptions, including associated phenotypes and optionally genotypes, into a central database
  (2) Have the option to unambiguously annotate cases using well-defined terms from e.g. SNOMED, ICD-10, HPO, FMA, etc
  (3) Automatically suggest terms (2) from free-form phenotype descriptions from clinical reports and pubmed literature
  (4) Have a search option that uses query expansion on the terms (e.g. including synonym, part-of, is-a relations) to find more related cases
  (5) Have a clustering/networking option to connect similar cases using some statistical measure and visualize that so it can be browsed when finding related cases
  (6) Have a fine grained 'facebook' like social networking and security mechanism to indicate with whom a doctor wants to share information
  (7) For UMCG only: have a way to add hospital information as phenotypic information on the cases (remember the abstract to HVP).
  It is also assumed that MOLGENIS can be used as overal framework, including authorization/authentication
  • (the idea of implementing it like facebook and assign each disease as a contact will not work so efficiently as if the contact is the doctor we hope the information will be filled more responsible otherwise we have to assign a disease to a doctor and persuade him to fill about the information . Besides , how will we connect to disease between them? In what terms are they related? Symptoms? well this scheme can be includes if we say that every doctor is a contact/entity which pick up his own contacts , and a disease is possible to be created as a contact /entity if needed.
• Implementation:
  Ad 1: expected is that the database model of Mark can be extends from the MOLGENIS for phenotypes, ​http://wwwdev.ebi.ac.uk/microarray-srv/pheno/molgenis.do
  Ad 2: expected is that the term selection feature can build on ontoCAT as well as Marks current backend
  Ad 3: expected is that the automated term matching can be build on Pergrine
  Ad 4: expected is that the query expansion needs to be written by ourselfs; previous experience suggests that use of an indexing service like Apache Lucene
  Ad 5: expected is that a clustering method can be incorporated to create a network of related cases based on Despoina's MSc thesis
  Ad 6: these security features are of general use in MOLGENIS and will be developed in collaboration with XGAP, AnimalDB etc.
  Ad 7: expected is that these issues can be addressed in collaboration with LifeLines and Mondriaan projects (remember the abstract to HVP).

1. SemOntoCat?  @ Ontocat : semantic jquery search : DEBUG/run simple appl with local data  does not work . 
   1. Added jquery conflict resolution found at​http://www.ibm.com/developerworks/web/library/wa-jquery1/
   2. Issue at ​http://github.com/loopj/jQuery-Tokenizing-Autocomplete-Plugin/issues/issue/17 Misc links 1. semantic search try simple jquery iterface from ​http://bassistance.de/jquery-plugins/jquery-plugin-autocomplete/
2. List of all : ​http://blog.yesnobox.com/post/JQuery-autocomplete-comparison.aspx
3. Jquery plugin ​http://www.noupe.com/ajax/45-fresh-out-of-the-oven-jquery-plugins.html