_PROBLEM CoEPrA-2006_Classification_003

_GROUP_NAME Wuju Li

_GROUP_MEMBERS
Hua Li
Yanyan Hou
Wuju Li

_ADDRESS
Center of Computational Biology, Beijing Institute of Basic Medical 
Sciences, Beijing, 100850, China


_MODELING_PROCEDURE
Description of Tclass classification system
In order to determine the class (-1/+1) of each member in the test 
dataset CoEPrA-2006_Classiification_003_Prediction_peptides.txt, Tclass 
classification system was used. Tclass system was developed in our center, 
which was originally written for the gene expression profile-based 
sample classification (Bioinformatics 2002, 18:325-326). In Tclass system, 
both Fisher and Naive Bayes prediction methods are integrated with the 
feature selection methods such as feature forward selection method. For 
the present task, the Naive Bayes method was used to determine the 
class (-1/+1) of each member in the test dataset.

Processes of classification
After input the training dataset and related class information (-1/+1) 
into the Tclass system 
(CoEPrA-2006_Classification_003_Calibration_Data.txt),
 the optimal three feature sets were found with the leave-one out 
cross-validation (LOOCV) as the object function. Prediction accuracy is 
0.8421. There are only three nona-peptides misclassified. Three optimal 
feature sets are as follows. 
The first set: 323, 753, 1926, 4524, 4825, 4878, and 5345
The second set: 323, 753, 1926, 4363, 4524, 4878, and 5345
The third set: 323, 753, 1926, 4008, 4524, 4878, and 5345
Where the number such as 323, 753, and etc. stand for the descriptors. 
For example, the 323 represents the 323rd descriptor.
In order to find the best feature set among the optimal three feature 
sets, the stability analysis was performed as follows for the particular 
feature set. First, the 89 nona-peptides in training dataset were 
divided into two groups randomly with partition ratio 75%, and the major 
part was used to construct the classifier. Second, the minor part was used 
to test the classifier. Third, above processes were repeated 1000 times 
and the average prediction accuracy from 1000 minor parts was taken as 
the stability index for the particular feature set. The stability 
indexes are 0.7977, 0.7991, and 0.7999 for the first, second, and third set 
respectively. Therefore, the third feature set (323, 753, 1926, 4008, 
4524, 4878, and 5345) was selected as the final best feature set, and 
the correspondent 1000 classifiers were taken as the final classifiers 
during the above stability analysis. The detail information for the 1000 
classifiers is provided in supplement 1.
In order to determine the class of a new nona-peptide, all 1000 
classifiers were used. If there are more than 500 classifiers to classify the 
nona-peptide into class -1, the nona-peptide is predicted into the 
class -1. Otherwise, the nona-peptide is predicted into the class 1.

Classification information for the training dataset
Based on the 1000 classifiers and the training dataset, we found that 
there were 21 nona-peptides misclassified (13th, 23rd, 25th, 39th, 42nd, 
45th, 47th, 60th, 62nd, 66th, 67th, 74th, 80th, 83rd, 87th, 90th, 93rd, 
105th, 109th, 110th, and 125th nona-peptide). Therefore, the number of 
nona-peptides for TP, TN, FP, and FN are 56, 56, 11, and 10 
respectively, which can be used to calculate sensitivity (Se), Specificity (Sp), 
and prediction accuracy as follows.
Se=TP / (TP+FN) = 84.85%
Sp=TN / (TN+FP) = 83.52%
Accuracy = (TP+TN) / (TP+TN+FP+FN) = 84.21%

Classification information for the test dataset
In test dataset, there are 133 nona-peptides. Based on the 1000 
classifiers, the classification information was provided in table 2. From 
table 2, we can see that there are 77 nona-peptides in class 1 and 56 
nona-peptides in class -1.

_PREDICTION
Obj_00001 +1
Obj_00002 +1
Obj_00003 -1
Obj_00004 +1
Obj_00005 +1
Obj_00006 -1
Obj_00007 -1
Obj_00008 -1
Obj_00009 -1
Obj_00010 -1
Obj_00011 +1
Obj_00012 -1
Obj_00013 -1
Obj_00014 +1
Obj_00015 -1
Obj_00016 -1
Obj_00017 +1
Obj_00018 +1
Obj_00019 +1
Obj_00020 +1
Obj_00021 -1
Obj_00022 +1
Obj_00023 -1
Obj_00024 +1
Obj_00025 -1
Obj_00026 +1
Obj_00027 -1
Obj_00028 +1
Obj_00029 +1
Obj_00030 +1
Obj_00031 +1
Obj_00032 +1
Obj_00033 +1
Obj_00034 -1
Obj_00035 -1
Obj_00036 +1
Obj_00037 -1
Obj_00038 +1
Obj_00039 -1
Obj_00040 +1
Obj_00041 +1
Obj_00042 -1
Obj_00043 +1
Obj_00044 +1
Obj_00045 +1
Obj_00046 +1
Obj_00047 -1
Obj_00048 -1
Obj_00049 +1
Obj_00050 +1
Obj_00051 +1
Obj_00052 +1
Obj_00053 -1
Obj_00054 +1
Obj_00055 +1
Obj_00056 -1
Obj_00057 +1
Obj_00058 +1
Obj_00059 +1
Obj_00060 -1
Obj_00061 -1
Obj_00062 -1
Obj_00063 +1
Obj_00064 +1
Obj_00065 +1
Obj_00066 -1
Obj_00067 -1
Obj_00068 +1
Obj_00069 +1
Obj_00070 +1
Obj_00071 +1
Obj_00072 +1
Obj_00073 +1
Obj_00074 -1
Obj_00075 +1
Obj_00076 +1
Obj_00077 +1
Obj_00078 -1
Obj_00079 -1
Obj_00080 +1
Obj_00081 +1
Obj_00082 -1
Obj_00083 -1
Obj_00084 +1
Obj_00085 -1
Obj_00086 +1
Obj_00087 -1
Obj_00088 -1
Obj_00089 +1
Obj_00090 +1
Obj_00091 +1
Obj_00092 -1
Obj_00093 +1
Obj_00094 +1
Obj_00095 -1
Obj_00096 +1
Obj_00097 -1
Obj_00098 -1
Obj_00099 -1
Obj_00100 +1
Obj_00101 -1
Obj_00102 +1
Obj_00103 -1
Obj_00104 +1
Obj_00105 +1
Obj_00106 +1
Obj_00107 +1
Obj_00108 +1
Obj_00109 +1
Obj_00110 -1
Obj_00111 -1
Obj_00112 +1
Obj_00113 -1
Obj_00114 -1
Obj_00115 +1
Obj_00116 +1
Obj_00117 +1
Obj_00118 -1
Obj_00119 +1
Obj_00120 +1
Obj_00121 +1
Obj_00122 -1
Obj_00123 +1
Obj_00124 -1
Obj_00125 -1
Obj_00126 -1
Obj_00127 +1
Obj_00128 -1
Obj_00129 -1
Obj_00130 +1
Obj_00131 -1
Obj_00132 -1
Obj_00133 +1