Uncovering the complexity of childhood undernutrition through strain-level analysis of the gut microbiome

Abstract

Background Undernutrition (UN) is a critical public health issue that threatens the lives of children under five in developing countries. While evidence indicates the crucial role of the gut microbiome (GM) in UN pathogenesis, the strain-level inspection and bacterial co-occurrence network investigation in the GM of UN children are lacking. Results This study examines the strain compositions of the GM in 61 undernutrition patients (UN group) and 36 healthy children (HC group) and explores the topological features of GM co-occurrence networks using a complex network strategy. The strain-level annotation reveals that the differentially enriched species between the UN and HC groups are due to discriminated strain compositions. For example, Prevotella copri is mainly composed of P. copri ASM1680343v1 and P. copri ASM345920v1 in the HC group, but it is composed of P. copri ASM346549v1 and P. copri ASM347465v1 in the UN group. In addition, the UN-risk model constructed at the strain level demonstrates higher accuracy (AUC = 0.810) than that at the species level (AUC = 0.743). With complex network analysis, we further discovered that the UN group had a more complex GM co-occurrence network, with more hub bacteria and a higher clustering coefficient but lower information transfer efficiencies. Moreover, the results at the strain level suggested the inaccurate and even false conclusions obtained from species level analysis. Conclusions Overall, this study highlights the importance of examining the GM at the strain level and investigating bacterial co-occurrence networks to advance our knowledge of UN pathogenesis.