ゲノム情報研究センター[CBRC]

2013年11月8日掲載

　2013年日本バイオインフォマティクス学会年会では約100件の発表ポスターの中から優秀な発表に対し、ポスター賞を授与しています。
　2013年10月29-31に東京・タワーホール船堀で開催された2013年年会・第2回生命医薬情報学連合大会にて、2013年日本バイオインフォマティクス学会年会のポスター賞として、独立行政法人産業技術総合研究所生命情報工学研究センター細胞システム解析チームの山田和範研究員と富井健太郎チーム長による以下の発表が選出されました。

ポスター概要

• タイトル:Development of a novel amino acid substitution matrix for remote homology detection
• 著者:Kazunori Yamada and Kentaro Tomii

　Amino acid sequence comparison method is the most fundamental tool for a wide spectrum of biological studies including protein structure and function prediction. To improve its performance, optimizing an amino acid substitution matrix is indispensable. Although a lot of amino acid substitution matrices have been developed, it has not been well understood which one is the best for similarity search especially for remote homology detection.
　In this study, we tackled to develop a novel matrix capable of detecting remote homology by collecting the information of existing matrices and condensing it. At first, we conducted a principal component analysis (PCA) with nine of typical existing matrices and acquired three PCA axes with high contribution ratio. In a PCA subspace constructed by these three axes, we searched the best detection performance matrix with SSEARCH for SCOP20 training dataset, which consisted of SCOP domains with 20% maximum pairwise sequence identity.
　We identified a matrix, designated as MIQS with the best detection performance and tested it on a validation dataset (half of SCOP20 dataset) and a test dataset, which consisted of CATH20 dataset ensured not to be shared homologous relationship with SCOP dataset by removing sequences related to entries in SCOP. As a result, MIQS showed improvement of detection performance compared to the existing matrices. In addition, the performance of MIQS with SSEARCH under actual use condition was superior to that of CS-BLAST, which was a novel high performance similarity search method without any amino acid matrix.
　As a conclusion, in this study we showed that by utilizing the PCA subspace based on the typical existing matrices, we obtained a novel matrix, MIQS, which improved the performance of homology detection. Using this matrix in combination with sophisticated comparison algorithm such as profile-profile comparison method, further improvement would be achieved.
　MIQS is available on SSEARCH at http://csas.cbrc.jp/Ssearch/.

写真：授賞式の様子

　表彰式では、日本バイオインフォマティクス学会長の浅井潔先生より賞状を授与され、連合大会参加者から多くの祝福と激励をいただきました。