Disease Diagnosis based on Various Pattern Frequency from Extracted Exons

Accurate disease diagnosis is an essential step to either prevent the disease or to personalize the treatment. Deoxyribonucleic Acid (DNA) has become an extremely useful source for diagnosing and predicting risks of many diseases. The disease diagnosis from DNA sequences is harmless but due to the huge size of sequences, the processing time is greatly affected. So, computational methods are proposed to speed up the diagnostic process and to improve diagnostic accuracy. Genes contain coding and non-coding regions called exons and introns. Genes act as instructions to make molecules called proteins from spliced exons. Usually Mutations arise spontaneously due to chemical instability of pyrimidine (Cytosine, Thymine, and Uracil) and purine (Adinine and Guanine) bases and also errors during DNA replication. The abnormality in the genetic composition due to mutations in exons causes a genetic disease in an individual. Based on the number of mutations, the complex diseases are generally categorized as normal state, pre-disease state or mutated state, and diseased state. To the best of our knowledge, none of the existing works focused on the exon pattern frequency based disease diagnosis. For better disease diagnosis, we created a disease database, which includes different patterns and corresponding range of frequencies for normal, mutated and diseased stages of complex diseases. In this work, we have proposed a novel multi-string pattern matching based computational method that leverages the Trie data structure for determining the frequency of patterns from exons. The proposed method compares the calculated frequencies with the created database ranges and decides the stages of disease. The comparison results bespeak that the proposed method has better speedup than the sequential and GPGPU based parallel Aho-corasick algorithm even for an increased number of patterns.