Classification Example

This example shows how to setup a classification experiment between Globins and Ser-Thr Kinases families. Note that this is not a challenging classification problem, and is shown here only to demonstrate to usage of scripts. Please see the publication for results of a more challenging classification experiment. The dataset and experiment setup is same as described in the Mining Globins Family page. The setup was the following set of proteins:

>> globals_set();
>> [ptns,fams]=data_prepare_experiment('experimentFile','families.globins.kinases.txt');
  1- 10 a.1.1.2 - (Globins)
 11- 28 a.1.1.* - (Globin-like)
 29- 38 d.144.1.7 - (Ser/Thr Kinases)
 39-227 _except_a.1.1.* - (except Globins)
228-424 _except_d.144.1.7 - (except Ser/Thr Kinases)

Now, we generate representative sets for both Globins and Kinases families, leaving one protein from each family out of the training.

>> mine_family_represent('globins_leaveoneout', 'ptns',ptns([1:9]), 'rand',ptns([39:227]),'recache',1);
>> mine_family_represent('kinases_leaveoneout', 'ptns',ptns([29:37]), 'rand',ptns([228:424]),'recache',1);

Next, we can test the proteins that were left out during the training phase and see which class they belong to:

>> class=classify_protein(ptns(10),{'globins_leaveoneout','kinases_leaveoneout'})

class =
     1

>> class=classify_protein(ptns(38),{'globins_leaveoneout','kinases_leaveoneout'})

class =
     2
>> 

Or, we could test all Globins and Kinases family members as follows:

>> for i=1:38; class(i)=classify_protein(ptns(i),{'globins_leaveoneout','kinases_leaveoneout'}); end; class

class =
  Columns 1 through 18 
     1     1     1     1     1     1     1     1     1     1     1     1     1     1     1     1     1     1

  Columns 19 through 36 
     1     1     1     1     1     1     1     1     1     1     2     2     2     2     2     2     2     2

  Columns 37 through 38 
     2     2

Note that the Globin-like proteins, some of which were not used in training (in the range 11-28) are also classified as Globins.

app/lfmpro/sample-classification.txt · Last modified: 2006/11/14 09:33 by Ahmet Sacan