Data mining and experimental approaches to identify combinations of natural herbs against bacterial infections

Abstract

Various studies have identified that natural herbs can be repurposed to treat infectious and bacterial diseases. The purpose of this study is first to test the medicinal value of five herbs including asafoetida, cumin, fenugreek, neem, and turmeric as single agent and in pairs using the bacterial zone of inhibition assay. Second, we used target and network analyses to predict the best combinations. We found that all the herbs as single agent were effective against bacterial infection in the following descending order of efficacy: cumin > turmeric > neem > fenugreek > asafoetida as compared to vehicle (ethanol) treated control. Among all the tested combinations the turmeric and fenugreek combination had the best efficacy in inhibiting the bacterial growth. Next to understand the mechanism of action and to predict the effective combinations among available herbs, we used a data mining and computational analysis approach. Using NPASS, BindingDB, and pathway analysis tools, we identified the bioactive compounds for each herb, then identified the targets for each bioactive compound, and then identified associated pathways for these targets. Then we measured the target/pathway overlap for each herb and identified that the most effective combinations were those which have non-overlapping targets/pathways. For example, we showed as a proof-of-concept that turmeric and fenugreek have the least overlapping targets/pathways and thus is most effective in inhibiting bacteria growth. Our approach is applicable to treat bacterial infections and other human diseases such as cancer. Overall, the computational prediction along with experimental validation can help identify novel combinations that have significant antibacterial activity and may help prevent drug-resistant microbial diseases in human and plants.