Identifying Trx-fold Protein

• Related papers:

1. Chang Wang, Stephen D. Scott, Jun Zhang, Qingping Tao, Dmitri E. Fomenko, and Vadim N. Gladyshev. A Study in Modeling Low-Conservation Protein Superfamilies. Technical report UNL-CSE-2004-0003, University of Nebraska, 2004. [pdf]  (With full description of the source of the data. The data set provided here is referred as "MIL (Motif-based alignment)". )
2. Qingping Tao, Stephen D. Scott and N. V. Vinodchandran. SVM-Based Generalized Multiple-Instance Learning via Approximate Box Counting. In Proceedings of the Twenty-First International Conference on Machine Learning (ICML 2004), pages 779-806, Banff, Alberta, Canada, July 2004. [pdf]
   

• Orignal sequences : [download files in fasta format]
  
• Data sets: [download]

1. Database file: primary.des.db
2. Specification file: primary.des.spec
3. Partitions for jackknife test
   a. positive examples: pos.svm
   b. negative examples: neg.svm.*
   

• File formats:

• Database file:
• Each line contains a bag;
• Each line / bag is in the format:  <label>    <number of points in the bag>    point_1 point_2 ... ...
• For example, a bag with label 1 and 3 points, (1,1,1), (2,2,2), (3,3,3) is written as

            1    3    1 1 1  2 2 2  3 3 3

• Specification file:

<number of classes>
<class label 1>    <class label 2>    ... ...
<dummy, no meaning>
<number of dimensions>
<minimum value for dimension 1>    <maximum value for dimension 1>
<minimum value for dimension 2>    <maximum value for dimension 2>
... ...

• *.svm* files:
• Each line refers to a bag in the database file;
• Each line is in the format: <label>    1:<index of the bag in the database file>
• The index is started with 0.