Geng Yang

A Dictionary-based Approach for Identifying Biomedical Concepts

In this research, we provided a dictionary-based approach for identifying biomedical concepts from the literature. The approach first crawled experimental corpus by E-utilities and built a concept ...

BCISeach: A Searching Platform of Breast Cancer Text Mining for Biomedical Literature

In this paper, we introduced a searching platform called BCISearch, a web-portal for the collection of molecular information linked to breast cancer using text mining technology including two types of information: static and dynamic information, involving with six categories entities and their relationships: protein, DNA, RNA, Cell-type, Cell-line, Virus. BCISearch could search 248997 Proteins, 71358 DNA, 7724 RNA, 58891 Cell-line, 871 Virus, 31698 Cell-type. The BCISearch Experimental approach is promising for develop biomedical text mining technology. The BCISearch would assist researcher to understand breast cancer etiology in genetic factors. The searching platform is available at http://210.28.186.168:8080/BCISearch.

Functional Analysis of Autism Candidate Genes Based on Comparative Genomics Analysis

In the post-genomics era, the rapid development of high-throughput technology makes data analysis become more and more important which could obtain some new biomedical knowledge, especially in understanding disease mechanism. In this work, we analyze the candidate genes related to autism by comparative genomics in two samples. We try to understand the autism disease pathology from molecular mechanism. We first select the confirmed autism susceptibility genes acting as positive sample, and the genes from biomedical literature related to autism by text mining technology acting as the unknown sample. By venn diagram analysis, the results obtain 25 autism susceptibility genes from the unknown sample. The results achieve some significant biomedical knowledge in the comparative functional analysis to the two samples, In GO analysis, we obtain that the two class of genes have some similar molecular functions including all kinds of binding functions. In the pathway analysis, VEGF signaling pathway and MAPK signaling pathway have significant enrichment about the two samples. The result also shows some genes between the two samples play a key role in the same signaling transduction pathway. It indicates that the functional analysis is helpful for candidate gene related to autism. This provides a way to study the disease from molecular mechanism.

RE-RANKING FOR PRIORITIZATION OF DISEASE-RELATED GENES

To explore the genetic complexity of disease, this paper focuses on the prioritization of susceptibility genes by re-ranking. This approach prioritizes disease-related genes based on the prior ranking. The genes in prior ranking are divided into seeds and candidates. And then these seed genes are used to prioritize candidates based on association rules. Experimental results show the approach is promising for finding new disease-related genes.

Mining miRNAs2target genes interactions from biomedical literature

This work describes an approach to mine miRNAs2target genes interactions related to breast cancer from literature. First, the work obtain 53 miRNAs and 2130 protein-coding genes regarding to the breast cancer, then using prior knowledge to extract the relationship between miRNAs and protein-coding genes, we achieved 2231 miRNAs2target interactions finally. We also constructed a miRNAs2target network, which could be used to clearly show the extracted miRNAs2target interactions.

A combined approach for the extraction of the multi-word and nested biomedical entity

Name entity recognition is the fundamental task in text mining area. This work focuses on the problems of multi-word and nested entity names. A combined approach is proposed for identifying multi-word and nested bio-entity names, which achieve an F-measure of 80.8% in extracting the total of bio-entity names and an F-measure of 82.2% aiming at nested entities. Experimental results show the combined approach is promising for developing text mining technology.