Neuro-fuzzy model of homocysteine metabolism

Abstract

In view of well-documented association of hyperhomocysteinaemia with a wide spectrum of diseases and higher incidence of vitamin deficiencies in Indians, we proposed a mathematical model to forecast the role of demographic and genetic variables in influencing homocysteinemetabolism and investigated the influence of life style modulations in controlling homocysteine levels. Total plasma homocysteine levels were measured in fasting samples using reverse phase HPLC. Multiple linear regression (MLR) and neuro-fuzzy models were developed. The MLR model explained 64% variability in homocysteine, while the neurofuzzy model showed higher accuracy in predicting homocysteine with a mean absolute error of 0.00002 μmol/L. Methylene tetrahydrofolate reductase (MTHFR) C677T, 5-methyltetrahydrofolate homocysteine methyltransferase (MTR) A2756G and 5- methyltetrahydrofolate homocysteine methyltransferase reductase (MTRR) A66G were shown to be positively associatiated with homocysteine, while nonvegetarian diet, serine hydroxymethyltransferase 1 (SHMT1) C1420T and TYMS 5'-UTR 28 bp tandem repeat exhibited negative association with homocysteine. The protective role of SHMT1 C1420T was attributed to more H-bonding interactions in the mutant modelled compared to the wild type, as shown through in silico analysis. To conclude, polymorphisms in genes regulating remethylation of homocysteine strongly influence homocysteine levels. The restoration of one-carbon homeostasis by SHMT1 C1420T or increased flux of folate towards remethylation due to TYMS 5'-UTR 28 bp tandem repeat or nonvegetarian diet can lower homocysteine levels.