Major Depressive Disorder is an acute-form of mental illness. It interferes in the personal life, education, eating and sleeping habits of a person affected by depression. The factors that cause depression are generally identified as environmental, genetic and other psychological reasons. Medications like anti-psychotic drug treatment and counseling are said to be responsive in controlling the mental condition for a short-term. But the current treatment methods are not effective for the patients, living with prolonged depression periods. Gene therapy gets its momentum on medical diagnostic procedures to treat the patients with handful strategies. Lipopolysaccharides, a kind of endotoxins presents in the outer membrane of gram-negative bacteria could cause potential threats to human body. In this work, the responsiveness of lipopolysaccharides simulated in blood of patients with depression over normal people is analyzed through their gene expressions. The samples are collected from Gene Expression Omnibus repository. A hybrid feature selection technique is proposed to select the biomarker genes of depression. Experimental results revealed the significant genes affected to Lipopolysaccharides simulation that discriminates the samples accurately. Machine Learning algorithms are employed to train and classify the data. This system finds 100% accurate classification of the normal and depression samples with the identified gene biomarkers. © IJSTR 2019.