http://bcb.ncat.edu/Features/

A Short Tutorial for FEPS (Feature Extraction from Protein Sequence) Webserver
***************************

This is a short tutorial that will walk you through the entire feature extraction process as performed in FEPS.

Scenario:
---------
Let's imagine that we have a classification problem in hand where we have five groups. Let's consider the protein sequences belonging to five protein groups as Group1.fasta, Group2.fasta,..., Group5.fasta. Here, the sequences of the proteins of each group are saved in a file that takes the name of the group. In such scenario, we usually need to extract numeric features from these sequences in order to use them as input in a supervised machine learning technique. FEPS provides 7 different feature types. In this tutorial, we will extract the feature using three feature type i) 'Amino Acid Composition' (without Feature type options) , ii) Autocorrelation Descriptors (Custom Geary Autocorrelation with ID, with Physicochemical Properties Option) and iii) Pseudo Amino Acid Composition (with Amino Acid Properties AAP option) 

i) Amino Acid Composition (Without Feature type options)

Fasta files:
------------
The five fasta files mentioned above are in the fasta directory of this tutorial. As discussed earlier, each file contains the protein sequences of a group. You can save these files in your hard disk as you will be asked to navigate to them.

Step 1: Choosing the Input File
+++++++++++++++++++++++++++++++
- With your browser go to http://bcb.ncat.edu/Features/
- Go to the Input: Protein fasta-formatted sequence file section and then Click 'Choose Files' button. 
- Navigate to the directory where you saved the fasta files and select all of them with ctrl+A and click on 'Open'. The selected files will be displayed on the 'The file list' area. 

Step 2: Choosing the Feature Type and Other related options
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
- From the 'Feature Types' radio button options choose the (feature type you want to calculate, in this case) 'Amino Acid Composition (AAC)'.
- From the dropdown menu select the default option 'Amino Acid composition'.
 (Amino acid composition doesn't require any option so all options fields are disabled.)

Step 3: Chossing the Output File Format
+++++++++++++++++++++++++++++++++++++++++
- From the 'Output file format' check the option of the output file format of your choice.
Let's say in this case, we choose first option 'Comma separated value (CSV) file'

Step 4: Providing Email address
++++++++++++++++++++++++++++++++++
- You can provide your email if you need the output files to be emailed to you. Otherwise you will be able to download the output as soon as the output files are available.

Step 5: Submit the job.
++++++++++++++++++++++
- Press the 'Submit' button to submit the job.
Step 6: Output Files
++++++++++++++++++++++++
The output page or the email will have five text files:
1- Sequence feature dataset file: It contains the feature numeric values. The features extracted from all files are saved in one file. Each column represent a feature and the last column contains the group label (numbered 0, 1, ... n-1 where n is the number of groups)
2- Sequence group names: The group alphanumeric description that correspond to the group labels. They are extracted from file names
3- Sequence feature title: The name of the chosen feature type
4- The feature type options: The options used with the feature type
5- Sequence feature names: The feature names or the feature column description


Once you downloaded the output files then you can use them with the software of your choice or you can modify them to suite your need.

 
ii) Feature Extraction based on Autocorrelation Descriptors (With Physicochemical Properties (ID) option)

Step 1: Choosing the Input File
--------------------------------
Same as Above

Step 2: Choosing the Feature Type and Other related options
----------------------------------------------------------
- From the 'Feature Types' radio button options choose the feature type 'Autocorrelation Descriptors'.
- Once you choose 'Autocorrelation Descriptors', you see the dropdown menu that shows 9 different options. Let's choose 'Custom Geary autocorrelation with ID' here. Please remember that any option with phrase 'ID' will enable choosing 544 physicochemical properties in selecting the 'Property'. Also, any option wiht phrase AAP will enable AA property dialog box. 
- Now, you can choose any of the 544 physicochemical properties as we have chosen 'ID'. Let's choose 'Residue Volume' option from the Physicochemical properties from 'Physicochemical properties drop down menu' of section 'Select a property or type an Amino Acid Index ID'.

Step 3: Choosing the Output file
-----------------------------------
- Chose the file format of your choice.

Step 4: Providing email address
-------------------------------
- Same as above

Step 5: Submit the job
----------------------
- Same as above

Step 6: Output Files
----------------------
- Same as above


iii) Feature Extraction based on Pseudo Amino Acid Composition (with Amino Acid Properties AAP option)

Step 1: Choosing the Input File
--------------------------------
Same as Above

Step 2: Choosing the Feature Type and Other related options
----------------------------------------------------------
- From the 'Feature Types' radio button options choose the feature type 'Pseudo Amino Acid Composition'.
- Once you choose 'Pseudo Amino Acid Composition' descriptors, you see the dropdown menu that shows various options. Let's choose 'Type 1 Pseudo amino acid composition with AAP' here. Please remember that any option with phrase 'AAP' will enable the 'AA property' dialog box. Here, you have an option to choose your own AAP (Amino Acid Properties file) by clicking the "Enter your AAP" Dialog box,  if you do so then you can enter your own indices for 20 different amino acids. Or you can choose any of the AA Property provided in the 'Dropdown menu' under 'AA Property'. Let's say we choose Hydrophobicity.
 
Step 3: Choosing the Output file
-----------------------------------
- Chose the file format of your choice.

Step 4: Providing email address
-------------------------------
- Same as above

Step 5: Submit the job
----------------------
- Same as above

Step 6: Output Files
----------------------
- Same as above

Contact: Hamid Ismail, hdismail@ncat.edu or Dr. Dukka KC dbkc@ncat.edu for any questions or comments. 




