
 ____                                       ____    ____    ____      
/\  _`\   __         __                    /\  _`\ /\  _`\ /\  _`\    
\ \,\L\_\/\_\  _ __ /\_\  __  __    ____   \ \ \L\ \ \,\L\_\ \ \L\ \  
 \/_\__ \\/\ \/\`'__\/\ \/\ \/\ \  /',__\   \ \ ,__/\/_\__ \\ \  _ <' 
   /\ \L\ \ \ \ \ \/ \ \ \ \ \_\ \/\__, `\   \ \ \/   /\ \L\ \ \ \L\ \
   \ `\____\ \_\ \_\  \ \_\ \____/\/\____/    \ \_\   \ `\____\ \____/
    \/_____/\/_/\/_/   \/_/\/___/  \/___/      \/_/    \/_____/\/___/ 


===============================Quick Start===============================

1) To run Sirius PSB, simply go to the folder where Sirius.jar is saved in the command prompt and type
	
   "java -Xmx1024m -jar Sirius.jar"
	
   note: computer would need to have at least 1024MB of RAM for above command. 
	However, Sirius PSB would also work using less RAM as it depends on size of datasets and program settings.
	

================================New Users================================

1) It is recommended that new users read through manuscript.pdf in this folder for an overview of the program.

2) Also, new users are encouraged to run through Example1(ChloroP).pdf and/or 
	Example2(Arabidopsis).pdf in order to have an idea of the software's capabilities.


=============================Advanced Users==============================

1) Advanced users who would like to reproduce the results stated in manuscript.pdf, please read the following.
	
	I) Subcellular Localization of Proteins
	----------------------------------------
	a) Go to http://www.cbs.dtu.dk/services/TargetP/datasets/datasets.php
	b) Download all the fasta files
	
	note: for nuclear sequences and Cytosolic sequences under plant, download the 54 sequences and 108 sequences 
	respectively.
	Then combine them as under other sequences. 
	Likewise, for the non-plant sequences of Nuclear and Cytosolic, combine them under other sequences.
	
	c) Then follow the instructions of manuscript.pdf to get results stated.


	II) Recognition of Polyadenylation Sites from Arabidopsis Genomic Sequences
	----------------------------------------------------------------------------
	a) Download "GIW2007-v2.zip" from http://www.comp.nus.edu.sg/~wongls/projects/dnafeatures/giw07-supplement/
	b) Then follow the instructions of Example2(Arabidopsis).pdf
			
		Except at step1 of Trainer: 
		Under the section of "Split all -ve Sequences into 3 Dataset", choose one of the following settings
		1) To obtain results of control sequences as 5UTR - Set "Dataset 3 (Use as Test Set): Seq# 3831 To 4118"
		2) To obtain results of control sequences as Coding - Set "Dataset 3 (Use as Test Set): Seq# 4119 To 4619"
		3) To Obtain results of control sequences as Intronic - Set "Dataset 3 (Use as Test Set): Seq# 4620 To 5146"					
		Except at step2 of Trainer:
		Load genetic_algorithm_144.features

	III) Shaping the Tubular Endoplasmic Reticulum in Saccharomyces Cerevisiae
	---------------------------------------------------------------------------
	a) Simply follow the instructions of Example3(NNSearcher).pdf


2)Trainer (Train Prediction Model)
	a) This application mainly allows users to train classifiers either using the 3 step methodology (feature generation, feature selection and feature integration) or the 4 step methology (feature generation, feature selection, feature integration and cascade classification)
	b) Users can also do cross-validation or blind set testing to get the trained classifier accuracy.
	c) Things that are possible to be saved/loaded:
		At Step1: Step 1 settings - ease of reloading the same settings.
		At Step2: Step 2 settings - ease of reloading the same set of features.
		At Step4: Score File - ease of future reference
		Classifiers - to allow trained classifier to be used in future.
	d) It also include a genetic algorithm to generate features from data if users are unsure of what features to use. The use of this genetic algorithm depends on the user's imagination. We have yet to exhaustively test the way to produce the best results. However, our method is to would do multiple runs of the genetic algorithm, then combine the features generated in different runs to get our good results. Users can use the "Feature File Manipulation" button in "Step2" for easy combination of features from different genetic algorithm runs. 


3) Predictor (Use Prediction Model)
	a) This application basically allows trained classifier to be run on fasta files.
	b) It can also be used to look at the score file of fasta files which were previously ran.
	c) It has a graphical view allowing for easy viewing of the prediction scores at different positions.
	d) It even allows users to set the classifiers to make predictions only when specified motifs are encountered in the sequence.
		

4) NNSearch (Nearest Neighbor Search)
	a) This application allows users to search for proteins not by sequence similarities, but more importantly by defining features such as amino acid composition and physiochemical properties such as hydrophobicity, charge and mass.
	b) It also has the ability for users to insert constraints to allow for more refined searches.

5) Misc (Pre/Post Prediction Utilities)
	a) This application assists users in building a classifier by providing applications to carry out preliminary and post prediction analysis.
	b) Some of these applications are:
		i) Visualization tools - allow users to identify trends in the sequences
		ii) Generation of position weight matrix - potentially helps user to identify conserved positions in the sequences
		iii) Sequence redundancy reduction - essential step for a newly collected set of sequences to avoid bias in training and testing of classifiers
		iv) Post prediction analysis - reqyuired when users are interested in combining predictions of several classifiers


==================Questions, feedback and bugs report===================

Please email kohchuanhock@nus.edu.sg
