| Summary: | 碩士 === 國立臺灣大學 === 資訊工程學研究所 === 94 === This dissertation provides a systematic methodologies for the information retrieval from digital media incorporating the knowledge from both biology and computer science. Here the work is proceeded in the genomic level to find out the evolutionary trends of enzyme proteins. Besides this, this thesis also illustrates some biological findings of how the enzymes can be related with each other through co-occurrences in the digital literatures.
Here, we developed a method of pseudo-reverse mechanism to compare the behaviour the enzyme proteins with the existing standard concepts. Our work is based on the strong assumption from the evolutionary theory, about the rates of nucleotide substitutions; we use this in the pseudo-reverse approach to verify how far it can be justified. We employed here the standard model of Nei and Gojobori in a generalized form for determining the nucleotide substitutions and Jukes and Cantor''s model for finding out their rates. We also embedded the comparative genomics in this model to calculate the lineages among the species like human, mouse and rat for these enzyme
proteins. We predicted from this study that the mutation for the enzymes are comparatively slower than ordinary proteins and the time of divergence for these enzymes with human and mouse or rat is almost five times more, around 400 Million years.
In the Appendix part of this thesis, we described the study on the enzyme co-citations. This involves automated extraction of explicit and implicit biomedical knowledge of the existing works on enzymes from the digital documents. We have presented here the work on a small scale data-set so as to have an overall idea of the availability of these enzymes on a generic digital library like CiteSeer. We created enzyme-to-enzyme co-citation network from digital documents from 4950 pairs of enzymes. This study emphasizes three basic statuses of the enzyme studies on the generic database like CiteSeer -- some are well established, some are half cooked and others still now unknown and unclear. Our goal is very simple and it mainly responsible to focus on two enzyme relation in a document. We validated the concepts of this work with the related references and found that this approach can find ways to detect diseases, which are caused or cured by certain enzymes. Even it can help to get the detail underlying molecular reactions about enzymes from the literatures.
|
|---|
