COMP 6703 Project A GUI Interface to the Gene Microarray Data Analysis Program SigMotif. Student: Ye Luo (u3972395) Clients: Professor Susan Wilson and.

COMP 6703 Project

A GUI Interface to the Gene Microarray Data Analysis

Program SigMotif.

Student: Ye Luo (u3972395)Clients: Professor Susan Wilson and Dr Kristine HardySupervisor: Dr Peter StrazdinsTime: 2006 Semester One

Background

• 1. Microarray experiments generate huge data sets

Microarray is an orderly arrangement of gene samples.

Gene Microarray• "1" " Id" "Score" "Ratio" "Human" "Mouse"

"Rat" "Name"• "2" "305677_A" "0.754102827479603"

"1.28401227454259" "ENSESTG00000047336" "NA""NA" "NA"

• "3" "813714_A" "0.692246894535161""1.59285273685750" "ENSG00000003402""ENSMUSG00000026031" "ENSRNOG00000012473""CFLAR"

• "4" "230261_A" "0.665698686323784""1.61538048354537" "ENSG00000006451""ENSMUSG00000008859" "ENSRNOG00000013454""RALA"

• "5" "488061_A" "0.118463671125342""3.81975136034479" "ENSG00000010438""ENSMUSG00000036938" "ENSRNOG00000013655""PRSS3"

• 2. No available software that combines the visualization and analysis of microarray data with that of biochemical networks (regulatory, interaction, metabolic).

SigMotif Description

A gene microarray data analysis program is developed by

an escience student Kristine Hardy last year.

It is built on the best of open-source analysis software R.

why use R ? R is an open source statistical computing environment

controlled by an interpreted programming language.

R provides good method and functions for easy

manipulation of statistical tests and graphs.

Some extension packages such as widgetTools provide

GUI interface to R program

SigMotif System Component

R

RWidgetTools

SigMotif

Systems Component

R a programming environment for dataanalysis and graphics

R Widget Tools an useful package for Rthat develops a GUI interface.

SigMotif a gene microarray data analysisprogram

Input and Output data of SigMotif

INPUT 1. List of motifs with description 2. Rat motif lists (Up, Down, Unchanged) 3. Human motif lists (Up, Down, Unchanged) 4. Mouse motif lists (Up, Down, Unchanged) 5. Significance/Homology lists (Up, Down, Unchanged) 6. Significance/Homology lists for comparison treatment

OUTPUT

1. Lists of significant motifs for different conditions

2. Graph of the significance values of given motifs for different conditions

3. Graphs of the conservation values of given motifs for different conditions

4. Lists of the motifs with the highest conservation values for a comparison treatment

Drawbacks in SigMotif GUI Interface

SigMotif program is unstable. SigMotif program often crashes. The main reason: GUI Interface is not robust. Other problems: . GUI Interface is not flexible and aesthetic. . Layout of GUI interface objects is unreasonable. . GUI Interface does not agree with users’ common

customs

New GUI Requirement

. Keeping the current design style

GUI Interface is composed of several dialogues

. Reasonable arrangement of different widget objects on GUI panel

. Important Improvements

The redundant dialogues

New GUI requirement cons

The redundant widget objects

The fixed buttons

Error checking

Feedback message

The length of text entry

……

New GUI Interface Modeling

• There are two different methods to be used to develop new GUI Interface

1. WXPython under Python environment

2. R Tcl/Tk under R environment

SigMotif with WXPython

Python WXPython

RPy

SigMotif

System Component1. Python . a dynamic object-oriented programming language . strong support for integration with other language and tools

2. WXPython . a GUI toolkit for the Python programming language

3. RPy . a Python interface to the R programming language . manage all kinds of R objects and execute arbitrary R functions

SigMotif with R Tcl/Tk

R

R Tcl/Tk

SigMotif

System Component

1. R

. a statistical computing environment

. provide good method and functions for easy manipulation of statistical tests and graphs

2. R Tcl/Tk

. a extension package connecting R to graphical toolboxes that can be used for developing menus and dialog boxes

The flowchart of the New GUI interface

Step0 Make or Load experiment

Step1 Load gene ormotif lists

Step4 the Menu

Step3 Save Exp Data

Step5 Get Exp

Step6 List significantmotifs

Step2 Make homologtables

Step10 Make Network

Step7 Graph significantmotifs

Step8 List conserved motifs

Step9 Graph conserved motifs

Program Implementation

• wxPython under Python environment Software Packages Installation

. Python . wxPython . R . RPy . Extensions including NumPy and Mark

Hommand’s

Important the software version number matching

Program Implementation

• R Tcl/Tk under R environment

Software packages installation . R . R Tcl/Tk

The R Tcl/Tk can be built under the R library directory

Basic steps of building a GUI dialogue

• 1. Load the R Tcl/Tk package under R environment

• 2. Define an R environment object for saving all intermediate data files and final output data files

• 3. Create a new top-level window• 4. Define widget objects• 5. Use command tkgrid to specify the layout• 6. Bind events to functions and generate events

Implementation Problems

New program version with wxPython1. The appearance is beautiful2. calling complicated R functions from

python is difficult.3. the program is unstable different input data files stop and even crash!4. RPy object conversion

Testing

Expected Output Data. Using the standard input data files provided by former designer and client Dr. K Hardy. There are totally 13 different .txt files. The expected Output Data list of significant motifs list of conserved motifs graph of significant motifs graph of conserved motifs

Robustness Performance

• 1. only one single input data lists loading dialogue

• 2. freely transform among different GUI dialogues

• 3. delete the redundant widget objects• 4. improved layout of widget objects on

top-level window• 5. feedback message• 6. environment variable

Conclusion and Future Work

1. wxPython is a good toolkit for developing GUI interface 2. How to resolve the interaction between R and Python? RPy???3. R Tcl/Tk is good. But a huge multi-line R program the layout method4. Requirement analysis Step10 GUI is too complicated too many widget objects text label of the widget objects