Changeset 1064


Ignore:
Timestamp:
May 14, 2009, 12:26:03 PM (12 years ago)
Author:
Jari Häkkinen
Message:

Addresses #118. Calculate target with median instead of mean.

Location:
plugins/base2/net.sf.basedb.normalizers/trunk
Files:
2 edited

Legend:

Unmodified
Added
Removed
  • plugins/base2/net.sf.basedb.normalizers/trunk/README

    r1051 r1064  
    5050== qQuantile normalization ==
    5151
    52 Write me.
     52qQN geometric mean non-logged values!
     53qQN must have positive numbers!
     54
     55The assays are normed against a selectable sub-set of the assays
     56... If a probe has no well-defined measurement (i.e., no assay in
     57the reference has a well defined value for a probe) it is simply
     58ignored from the target distribution.
     59
     60Ja, jag tankte att man far justera q till maximalt antal icke nan ...
     61men som sagt an sa lange ar q==100.
     62
     63Den begränsning som finns nu är att det måste finnas lagom många
     64väldefinierade mätvärden per assay (några hundra per assay) annars
     65kraschar nog programmet. Jag kommer att lösa detta genom att välja
     66antalet bins i distributionsberäkningen som #bins=max(100,N_i/10)
     67i=1...#assays och kräva att #bins är minst 10. Kravet är alltså att
     68varje assay måste ha minst 100 väldefinierade punkter. Om inte kravet
     69är uppfyllt stannar programmet med ett någorlunda trevligt meddelande.
     70
     71In q-quantile normalization each assay data is sorted in ascending
     72expression value order and added to a matrix as columns. The matrix
     73rows will contain mixed probes (also known as reporters or genes)
     74decided by their rank. For each row in the matrix, the expression
     75values are replaced with the row average value. Finally, each assay is
     76reordered into its original order to retain a standard expression
     77matrix were each row represents one probe. Assays are not mixed.
     78
     79Background subtraction and proper filtration should be done on the
     80bioassay set before running this plug-in. The bioassay set must not
     81contain any missing values.
    5382
    5483The qQuantile normalization is inspired by the 'Cubic Spline'
  • plugins/base2/net.sf.basedb.normalizers/trunk/src/net/sf/basedb/plugins/qQuantileNormalization.java

    r1043 r1064  
    8181        "Background subtraction and proper filtration have to be done on the " +
    8282        "bioassay set before running this plug-in.\n\n" +
    83         "1-channel or 2-channel are supported\n\n" +
     83        "Only 1-channel data is supported\n\n" +
    8484        "This plug-in is part of the BASE normalizers package.",
    8585        Normalizations.getVersion(),
     
    215215                     "--assay-data", assaydata,
    216216                     "--in-data", data_before,
    217                      "--out-data", data_after };
     217                     "--out-data", data_after,
     218                     "--median" };
    218219    execute(cmd);
    219220  }
Note: See TracChangeset for help on using the changeset viewer.