Introduction

 

SOFTWARE

 

BioDesigner contains two parts, server part and client part. The codes for server part are written in java programming language and the codes for client part are written in C++ programming language. The GUI is designed with Qt which is based on C++. As a result, all of them can be compiled across various platforms. The program has been built on Windows, Linux and Mac operating systems.

The server part is used to build the database that contains the information of existing devices, and make analysis on these information. The client part consists of GUI and analysis of users input. Users input there information through GUI, and client codes deal with the input and send requirements to the server. Server receives the requirements from client, and makes analysis based on the database. Then, it returns information to client. Client deals with the information and visualizes it.

OThe database is stored in server, which would simplify the installation. Since the server is more powerful, the algorithms for recommendation are executed on server, which will reduce the analysis time.

 

METHODOLOGY

In order to accomplish the function of BioDesigner, such as recommending biobricks, we adapt various recommendation algorithms to achieve the goal. These algorithms include Collaborative filtering and Association rules.

1. Collaborative filtering:

a. The collaborative filtering based on the content: Given a biobrick, Biodesigner finds a similar biobrick to recommend according to the sequence of biobrick.

b. The collaborative filtering based on the user: By storing the biobricks users have been used, Biodesigner statistics users’ habit to recommend the higher frequency biobrick.

2. Association rules: Based on existing device chains, implied relationship, including relationship between two successive biobricks, three successive biobricks, four successive biobirkcs and five successive biobricks, are obtained. Biodesigner uses these relationships to recommend the biobrick over the minimum confidence.

3. By the link between parts to recommend (recommended homologous part).

To meet the need of the analysis and recommendation, our team innovatively improved most of algorithms and methods to make the simulation more accurate and more efficient.

 

DATABASE

In order to support Biodesigner, we build a database that contains the systems designed by the teams of previous years. About 2500 device chains are collected and stored in our database. We also download the biobrick information contained by these chains from Registry of Standard Biological Parts and stored these information in our database.