It is very important to select a small set of relevant features from a high dimensional data set and useful to design either an effective classification or prediction model. This procedure involves a series of estimations of the relationship between each pair of variables and between each variable and class labels. Mutual information is widely used to estimate these relationships. However, alternative strategies may be useful to estimate the mutual information with continuous or hybrid data. In this study, we attempt to evaluate the difference between the selection strategies involved with mutual information and fuzzy mutual information. The results indicate that using fuzzy mutual information is more helpful to obtain more stable feature sets and more accurate estimations of the relationship between two variables.

BACKGROUND: Recombinant protein production is a useful biotechnology to produce a large quantity of highly soluble proteins. Currently, the most widely used production system is to fuse a target protein into different vectors in Escherichia coli (E. coli). However, the production efficacy of different vectors varies for different target proteins. Trial-and-error is still the common practice to find out the efficacy of a vector for a given target protein. Previous studies are limited in that they assumed that proteins would be over-expressed and focused only on the solubility of expressed proteins. In fact, many pairings of vectors and proteins result in no expression. RESULTS: In this study, we applied machine learning to train prediction models to predict whether a pairing of vector-protein will express or not express in E. coli. For expressed cases, the models further predict whether the expressed proteins would be soluble. We collected a set of real cases from the clients of our recombinant protein production core facility, where six different vectors were designed and studied. This set of cases is used in both training and evaluation of our models. We evaluate three different models based on the support vector machines (SVM) and their ensembles. Unlike many previous works, these models consider the sequence of the target protein as well as the sequence of the whole fusion vector as the features. We show that a model that classifies a case into one of the three classes (no expression, inclusion body and soluble) outperforms a model that considers the nested structure of the three classes, while a model that can take advantage of the hierarchical structure of the three classes performs slight worse but comparably to the best model. Meanwhile, compared to previous works, we show that the prediction accuracy of our best method still performs the best. Lastly, we briefly present two methods to use the trained model in the design of the recombinant protein production systems to improve the chance of high soluble protein production. CONCLUSION: In this paper, we show that a machine learning approach to the prediction of the efficacy of a vector for a target protein in a recombinant protein production system is promising and may compliment traditional knowledge-driven study of the efficacy. We will release our program to share with other labs in the public domain when this paper is published.

BACKGROUND: Schizophrenia is a complex disease with multiple factors contributing to its pathogenesis. In addition to environmental factors, genetic factors may also increase susceptibility. In other words, schizophrenia is a highly heritable disease. Some candidate genes have been deduced on the basis of their known function with others found on the basis of chromosomal location. Individuals with multiple candidate genes may have increased risk. However it is not clear what kind of gene combinations may produce the disease phenotype. Their collective effect remains to be studied. RESULTS: Most pathways except metabolic pathways are rich in protein-protein interactions (PPIs). Thus, the PPI network contains pathway information, even though the upstream-downstream relation of PPI is yet to be explored. Here we have constructed a PPI sub-network by extracting the nearest neighbour of the 36 reported candidate genes described in the literature. Although these candidate genes were discovered by different approaches, most of the proteins formed a cluster. Two major protein interaction modules were identified on the basis of the pairwise distance among the proteins in this sub-network. The large and small clusters might play roles in synaptic transmission and signal transduction, respectively, based on gene ontology annotation. The protein interactions in the synaptic transmission cluster were used to explain the interaction between the NRG1 and CACNG2 genes, which was found by both linkage and association studies. This working hypothesis is supported by the co-expression analysis based on public microarray gene expression. CONCLUSION: On the basis of the protein interaction network, it appears that the NRG1-triggered NMDAR protein internalization and the CACNG2 mediated AMPA receptor recruiting may act together in the glutamatergic signalling process. Since both the NMDA and AMPA receptors are calcium channels, this process may regulate the influx of Ca2+. Reducing the cation influx might be one of the disease mechanisms for schizophrenia. This PPI network analysis approach combined with the support from co-expression analysis may provide an efficient way to propose pathogenetic mechanisms for various highly heritable diseases.

The 2008 annual conference of the Asia Pacific Bioinformatics Network (APBioNet), Asia's oldest bioinformatics organisation set up in 1998, was organized as the 7th International Conference on Bioinformatics (InCoB), jointly with the Bioinformatics and Systems Biology in Taiwan (BIT 2008) Conference, Oct. 20-23, 2008 at Taipei, Taiwan. Besides bringing together scientists from the field of bioinformatics in this region, InCoB is actively involving researchers from the area of systems biology, to facilitate greater synergy between these two groups. Marking the 10th Anniversary of APBioNet, this InCoB 2008 meeting followed on from a series of successful annual events in Bangkok (Thailand), Penang (Malaysia), Auckland (New Zealand), Busan (South Korea), New Delhi (India) and Hong Kong. Additionally, tutorials and the Workshop on Education in Bioinformatics and Computational Biology (WEBCB) immediately prior to the 20th Federation of Asian and Oceanian Biochemists and Molecular Biologists (FAOBMB) Taipei Conference provided ample opportunity for inducting mainstream biochemists and molecular biologists from the region into a greater level of awareness of the importance of bioinformatics in their craft. In this editorial, we provide a brief overview of the peer-reviewed manuscripts accepted for publication herein, grouped into thematic areas. As the regional research expertise in bioinformatics matures, the papers fall into thematic areas, illustrating the specific contributions made by APBioNet to global bioinformatics efforts.

Summary form only given. Traditional domain database only provides the sequence related information, such as the consensus sequence or a position-specific scoring matrix. The so-called integrated domain databases, such as InterPro or CDD (conserved domain database), only integrates the information from different types of domain databases. The functional information, such as alternative splicing, tissue, protein interactions, etc., are usually missing. Therefore, it is important to have an integrated protein domain database, which will create a user-centric environment for biologists. The advantage of creating an integrated database for domains, instead of genes, is to discover the relations within a gene family. We have used death domain as an example to explain how to integrate the genome, transcriptome, and proteome information together. In this user-centric bioinformatics environment, the users may use information to make observations and even to create hypothesis for biomedical research. In other words, information integration is essential for information-driven biomedical research. For example, by integrating the alternative splicing and the protein domain information, a protein variant that may show inhibitory effect on signal transduction can be observed. By integrating tissue information with alternative splicing, protein variant that will be expressed in a given type of tissue might be revealed. Besides, by integrating tissue and histology information with pathway, the difference of a pathway in normal versus tumor cells would be clear. This system can be readily be applied to other protein domains. Thus, we name the whole program package "integrated domain resource (IDR)". The ultimate goal for this database is to integrate medical information into the system, so the relation of genotype and phenotype can be established in the future.

UNLABELLED: The positional candidate gene approach accelerates the discovery of genes involved in disease. However, the properties of such disease genes are very diverse and the sample size of known disease genes is too small and does not warrant success by the use of a machine-learning approach. A user-defined scoring system may thus help to determine the priority of candidate genes. Spinocerebellar ataxia (SCA) is a good model to test this approach because most SCA subtypes are caused by an expansion of short tandem repeats (STRs). The SCA db is a candidate gene database for SCA, which collected 3185 genes for 17 types of SCA. Those SCA subtypes that have known disease genes can be used as positive controls to optimize the parameters. The users may browse the candidate genes of a given SCA subtype by using the default parameters. The known disease genes were found to be the top three candidates using the default parameters. Alternatively, the users may score the candidate genes by changing the weight or the scores on the basis of their own working hypothesis. AVAILABILITY: This database is available at http://ymbc.ym.edu.tw/sca/