Automatic Teaching–Learning-Based Optimization: A Novel Clustering Method for Gene Functional Enrichments - Computational Intelligence Techniques for Comparative Genomics - page 18

silhouettes (SIL) [ 7 ], Davies and Bouldin (DB) [ 8 ], and Chou (CS) [ 9 ] measures,
primarily nds the best partitioning in the underlying data.

This paper impersonate k-means clustering algorithm, procedures for automatic
clustering, CVIs, visualization, and elitism techniques into TLBO. The objective of
the novel AutoTLBO algorithm was to cluster the Saccharomyces cerevisiae cat-
egorized microarray datasets, and the expected multiple outcome was to attain
optimal number of automatic clusters, mean values of CVIs, dendrograms and
cluster pro

les of co-expressed genes. These outcomes are assumed as summative
assignment-I and is shown as Experiment 1 in Sect. 5 . The obtained cluster pro les

of AutoTLBO are used as inputs into Bioinformatics tools FatiGO for rst opinion

and database for annotation, visualization, and integrated discovery (DAVID) for
second opinion. This veri cation procedure is named as summative assignment-II,

primarily used to re-validate the results given by this novel AutoTLBO. Figure 1
unveils a broad road map of the proposed work in this paper. The input is fed into
the tool in such a manner that the

rst list holds the gene-IDs of one of the cluster
and the other list holds the gene-IDs of all the remaining clusters generated by this
novel AutoTLBO algorithm. Two-stage preprocessing is imposed on the lists by
applying statistical techniques such as Fisher exact test and duplicate elimination.
Finally, these clean lists are used in gene ontology (GO) biological process to

nd

the signi cant terms, term annotations % per list, p-value, FDRs, enrichment scores,

etc. This entire set of test results of both the tools are publicized as Experiment 2 in
Sect. 5 . The re-validate techniques adopted in the tools manifest a positive sign that
the novel AutoTLBO is power-packed in obtaining optimal number of automatic
clusters and discrete gene cluster pro les. This silver lining absolutely ratify that the

novel algorithm proposed in this work can used for attaining gene functional
enrichments.
The rest of the paper is formed as follows. Section 2 exposes a basic background
to the theme concepts used by other researchers, TLBO, and its variations.

Fig. 1 Road map of AutoTLBO: a novel clustering method for gene functional enrichments