SkinBug can predict the metabolism and biotransformation of any molecule by the human skin microbiota. It can predict all possible reactions with reaction centers, corresponding enzymes, microbial species, and skin sites capable of metabolizing the input molecule.
Skin is the largest and most exposed organ of the human body. It harbours a diverse microbiota of microorganisms which have an enormous metabolic potential to carry out xenobiotic transformation of molecules. The skin microbiota comes in regular contact with pollutants, topical substances like skincare products & medical ointments etc., and can metabolize these. Based on current literature, there are ~1000 different species present in the skin microbiota and it is not possible to experimentally elucidate the metabolic contribution of this heterogeneous community towards a given molecule. Thus, we have developed a robust and efficient high throughput computational method. SkinBug exploits the promiscuity of metabolic enzymes (an enzyme can potentially metabolize multiple substrates which are structurally similar). SkinBug can predict multiple enzymes from multiple microbial species. This is referred to as a ‘multi-label multi-class classification’ problem. SkinBug exploits the molecular properties of substrates of metabolic enzymes along with their known reactions from ~900 microbial skin species belonging to ~19 different skin sites. Molecular properties of all substrates were translated into machine readable features (Descriptors, Fingerprints) and were used for preparing machine learning (Random Forest SRC - RFSRC) and deep learning models (Artificial Neural Network - ANN). We have achieved an accuracy of >80% which is very challenging to obtain for a multi-label multi-class classification problem. Considering the vital role of skin microbiota in human health, a deep understanding of the metabolic potential of this community along with their metabolic interaction with the different substances (medical, cosmetic, ayurvedic, etc.) applied to the skin will be very crucial for better healthcare. This tool helps further the field by providing insights for experimental studies which is a step closer towards the goal of personalized medicines and novel diagnostics and therapeutics.
SkinBug predicts EC number with reaction, microbial enzymes, skin microbes, and skin site for the metabolism of input molecule using the following steps: