Comparison of clustering methods for single-cell transcriptome analysis

• Item type: SIG Technical Reports(1)
• 公開日: 2017-09-19
• タイトル: Comparison of clustering methods for single-cell transcriptome analysis
• タイトル / 言語: en
• タイトル: Comparison of clustering methods for single-cell transcriptome analysis
• 言語: eng
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Center for iPS Cell Research and Application, Kyoto University
• 著者所属: Center for iPS Cell Research and Application, Kyoto University
• 著者所属: Center for iPS Cell Research and Application, Kyoto University
• 著者所属(英): en
• 著者所属(英): Center for iPS Cell Research and Application, Kyoto University
• 著者所属(英): en
• 著者所属(英): Center for iPS Cell Research and Application, Kyoto University
• 著者所属(英): en
• 著者所属(英): Center for iPS Cell Research and Application, Kyoto University
• 著者名: Yuji, Kozakura; Tomoya, Mori; Wataru, Fujibuchi
• 著者名(英): Yuji, Kozakura; Tomoya, Mori; Wataru, Fujibuchi
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Single-cell transcriptome can provide information of global gene expression pattern in individual cells, so that detailed cell type classification considering cellular heterogeneity becomes more important to analyze complex biological system. Here we surveyed 12 clustering methods utilized for single-cell transcriptome analysis. We evaluated the performance of each method using published data of 1,830 single-cell transcriptome obtained from SHOGoiN database. Each single-cell transcriptome data was labeled by source tissues (adipose tissue, blood, brain, early embryonic tissue, genitalium, muscle, pancreas, skin). In the evaluation, we chose two criteria; normalized mutual information (NMI) and the purity. After optimizing the parameters by NMI, we calculated the sum of the NMI and the purity. As a result, the combination of the DBSCAN algorithm and t-SNE clustering showed the best performance on our data set.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Single-cell transcriptome can provide information of global gene expression pattern in individual cells, so that detailed cell type classification considering cellular heterogeneity becomes more important to analyze complex biological system. Here we surveyed 12 clustering methods utilized for single-cell transcriptome analysis. We evaluated the performance of each method using published data of 1,830 single-cell transcriptome obtained from SHOGoiN database. Each single-cell transcriptome data was labeled by source tissues (adipose tissue, blood, brain, early embryonic tissue, genitalium, muscle, pancreas, skin). In the evaluation, we chose two criteria; normalized mutual information (NMI) and the purity. After optimizing the parameters by NMI, we calculated the sum of the NMI and the purity. As a result, the combination of the DBSCAN algorithm and t-SNE clustering showed the best performance on our data set.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AA12055912
• 書誌情報: 研究報告バイオ情報学（BIO）; 巻 2017-BIO-51, 号 2, p. 1-6, 発行日 2017-09-19
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8590
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会