The present article focuses on a systematic application of clustering algorithms (Fuzzy c-means (FCM) and Partitioning Around Mediod (PAM)) on gene expression data. We show a way of applying these algorithms to select some possible genes responsible for a particular disease. The genes those are severely over or under expressed in the allergen samples are identified. Two different techniques are applied on the same gene expression datasets containing both allergen samples and control samples. First technique uses clustering algorithms on expression values, followed by determining similarity/dissimilarity among control and disease clusters, and measuring the extent of over/under expression of genes from normal to disease condition. By the second technique, we apply clustering algorithms on fold values and measure the over/under expression of genes. By these two techniques we have identified several genes those have significantly changed their expression values for asthmatic condition, and have reported in the present article. Some of these observations are supported by some earlier investigations. Others have been stayed unnoticed so far, but may play crucial role in mediating the development of asthma..