Introduction to Bioinformatics Algorithm Laboratory

This group has wide range of research interest in computational molecular biology, including genome analysis,algorithm design and analysis.

  • Projects - a list of Faculty-led research projects

  • Resources - a list of biological resources


  • 2D image recognition and 3D image reconstruction of the Drosophila brain

    This study proposes a set of shape descriptors of a neuron image to match the 3D neurons in FlyCircuit called DescNeuro. A set of four descriptors is identified from the best combination of 23 potential shape based descriptors. Each 3D neuron is represented by a number of projected images from multiple viewpoints evenly distributed on a spherical surface. The shape-matching method is helpful for detecting putatively duplicated neurons if multiple pairs of projected images with high similarity and close location exist.

  • Digital pathological biopsy analysis platform

    We have a high resolution and high speed biopsy scanner, and the resolution can achieve 0.23gm/pixel. The core techniques are the proposed bioimage feature extraction from one HE-stained image and optimal feature selection methods based on our previously developed intelligent evolutionary algorithm. The result shows that the proposed pathology analysis system can effectively discriminate the HE stain images of biopsy. This 2-year research project is now financially supported by Delta Electronics Inc. with NTD$7,200,000.

  • Automatic neuron morphology quantification system

    We develop a mathematical modeling to identify 9 kinds of different phenotypes of neuron images by using IBCGA. The different phenotypes of neurons are regulated by 9 kinds of shRNA which has different genotypes.

  • Gene Regulatory Networks (GeNOSA & IRNet)

    The reconstruction of genome-wide gene regulatory networks is known as a challenge task in Systems Biology consisted of integration of biological domain knowledge, mathematical models for gene regulatory networks, and optimization of the proposed model with large parameters. We propose two methods, GeNOSA and IRNet, can reconstruct quantitative and robust gene regulatory networks (GRNs). By understanding the dynamics of these networks we can shed light on the mechanisms of diseases that occur when these cellular processes are dysregulated. To GRNs remains an indomitable challenge to reveal the quantitative relationships of transcription factors and genes under different conditions. From the quantitative GRNs established from microarray data by using the proposed framework, GeNOSA, ones can interpret the hidden information, such as the transcription factor activities and their effects to expression of down-stream genes. GeNOSA successfully established highly agreed GRNs compared to prior biological domain knowledge and experimentally validated by electrophoretic mobility shift assays and real-time PCR in E. coli. Using nonlinear model can capture true biological response, they difficultly inferring robust networks and with high false positive rate owing to the dataset has noise and a large number of genes but a small number of experimental data. In this study, we propose an incremental reconstruction network algorithm, IRNet, can determine regulations in each iteration, and the inheritable mechanism can inherit determined regulations to next iteration that gradually reduce the solution space. Finally, IRNet infers a robust network from a small number of replicated time-series expressions data.

  • Scoring card design for protein function prediction

    We cooperated with Prof. Huang (NCTU) to develop a simple and easily interpretable prediction method with informative sequence features to provide insights into protein crystallization. This study proposes an ensemble method, SCMCRYS, to predict protein crystallization, for which each classifier is built by using a scoring card method (SCM) with estimating propensity scores of p-collocated amino acid (AA) pairs (p = 0 for a dipeptide). The SCM classifier determines the crystallization of a sequence according to a weighted-sum score. The weights are the composition of the p-collocated AA pairs, and the propensity scores of these AA pairs are estimated using a statistic with optimization approach. SCMCRYS predicts the crystallization using a simple voting method from a number of SCM classifiers.

  • Neurocomputing-EEG motion sickness estimation system

    We cooperate with Prof. Ko (NCTU) and Brain Research Center (BRC) to develop an estimation system for motion sickness. We play an important role to choose the important brain area versus frequency by using our core technology. Based on IEA, the estimation system can predict the sickness level and the classifier can predict the sickness stage. The related works were published on IEEE 2012 International Joint Conference on Neural Networks (IJCNN 2012) and IEEE Symposium Series on Computational Intelligence 2013 (SSCI 2013).

  • Medical informatics-Medical and clinical data analysis system

    We cooperated with Prof. Huang (NCTU), Prof. Chen (NCTU), Changhua Christian Hospital, TMU Shuang Ho Hospital and Mackay Memorial Hospital to analyze the disease relationship by using National Health Insurance Research Database (NHIRD). In order to efficiently manipulate the big data of NHIRD, we build up a high-speed server which is comprised of professional database software and hardware for processing the database searching, comparing and analysis. Thereafter, the statistics of the interested research are incorporated, such as odds ratio, hazard ratio or survival curves.


  • Automatic neuron morphology quantification system
  • Gene Regulatory Networks (GeNOSA & IRNet)
  • Scoring card design for protein function prediction