* |
fzDBI: Density Based Imputation method for fuzzy cluster analysis of gene expression data.
|
|
We present fzDBI, a novel imputation method for fuzzy cluster analysis of gene expression microarray data with missing values;
Genes are clustered using the Fuzzy C-Means algorithm (FCM).
The fuzzy partition obtained is then used to create a
density-based fuzzy partition which is used with the FCM
fuzzy partition to estimate the missing values in the
dataset.
fzDBI outperformed other methods on both artificial and real datasets,
particularly on datasets with clusters that differed in size.
fzDBI is therefore appropriate for real-world datasets in which the data densities are not uniformly distributed.
|
|
|
Presentation
|
This will be presented at the Fourth International Conference on
Bioinformatics and Computational Biology (March 12-14, 2012, Las Vegas, Nervada USA),
scheduled on March 13th 2012 at 11:30am-12:00am.
|
|
Online demo
|
This online demo was developed using AJAX technology, PHP and C++ programming languages.
|
|
|
Artificial datasets
|
This package contains two artificial datasets generated using the method of Xu and Jordan (1996).
|
|
|
Real datasets
|
This package contains two datasets: iris, wine from UCI ML Repository
|
|
|
Gene expression datasets
|
This package contains two gene expression datasets: yeast, yeast-MIPS from Ka Yee Yeung
|
|
|
Additional gene expr. datasets
|
This package contains two gene expression datasets, RCNS (Wen et al., 1998) and Serum (Iyer et al., 1999); the benchmark reports
were added to the final version.
|
|
|
Artificial dataset results
|
This package contains the performance results of fzDBI and the other imputation methods on two artificial datasets.
This is reported in the Figures 2&3 of the manuscript.
|
|
|
Real dataset results
|
This package contains the performance results of fzDBI and the other imputation methods on the two datasets:
iris and wine. This is reported in the Figures 4&5 of the manuscript.
|
|
|
Gene expression dataset results
|
This package contains the performance results of fzDBI and the other imputation methods on the two gene expression
datasets: yeast andd yeast-MIPS. This is reported in the Figures 7&8 of the manuscript.
|
|
|
Additional gene expr. dataset results
|
This package contains the performance results of fzDBI and the other imputation methods on the two gene expression
datasets, RCNS and Serum, reported as in Figures 6&9 of the manuscript. An additional report, below is on the performance
of the algorithms on the Serum data normalized by genes.
|
|