Automatic Teaching–Learning-Based Optimization: A Novel Clustering Method for Gene Functional Enrichments - Computational Intelligence Techniques for Comparative Genomics - page 27

(a) (b)

(c) (d)

Fig. 3 Hierarchical expression prole of S. cerevisae categorized datasets. a cluster 4 of yeast238,
b cluster 4 of yeast384, c cluster 4 of yeast2885, d cluster 4 of yeast2946

The quality of the algorithm is inspected by applying with a huge quantity
dataset called yeast2885 with 2,885 gene-IDs and 19 dimensions. The ef ciency of

algorithm is exhibited by producing a value 5.4 as an optimal number of auto
clusters, low usage of CPU time, and low percentage of error rate. Optimal
threshold values are recorded at SIL and MIM indices to highlight the competence
of clustering accuracy. Hence, this outcome launches AutoTLBO as a mere and
potential immigrant in clustering microarray datasets. The core ideology of gen-
erating the automatic cluster when entrenched over yeast2885 is visualized as
Fig. 4 . This ve respective constellations of gene-IDs 1,422,

367, 877, 388, and 1,327 are prudent and consummated to the actual. Figure 3 cis
impinged with a heat map over the fourth cluster of yeast2885. The analysis from
the heat map [ 22 ] was the data matrix holds the color information of microarray
dataset along with numeric data. The red color is evidence for higher expression
level of the gene, whereas green indicates low expression level and black indicates
the absence of expression level. The perception was so bedazzle that the heat maps
generated by AutoTLBO have higher expression level since most of the cluster is
marked in red color.

gure justi es that the