Optimization Of Analysis Methodsand Its Application In Population-level Gut Microbiome Survey

BACKGROUD:Microbiome data analysis involves multiple steps,that each of them could affect the final results.Espeicially in big data analysis,questions like:Whether different datasets can be combined to do analysis?How to improve the stability and speed of microbiome data analysis is not fully answered.Meanwhile,Human gut microbiome is highly variable,which could affect the reliablity of microbiome studies and Population-level analysis,that encompasses a large variety of gut microbiome and host phenotypes is essential to comprehensively study the relationship between gut microbiome and host parameters.Such studies were rarely conducted in Eastern developing areas.Objectives:To optimise microbiome analysis methods,regarding its stability and speed,and based on those methods,to conduct a survey of gut microbiome in Guangdong province on population-level.Methods:Part 1 1.By comparing the same varable 16S rRNA gene tags but generated from different primer sets,we observed how experimental details can affect microbiome analysis;2.By using the same dataset that reported the non-overlapping rarefaction curves,we illustrate the stability of OTUs clustering and developed a relatively stable clustering methods;3.By parallelizing de novo clustering methods,we developed Subsampled open reference OTU picking pipeline that increases the speed of de novo clustering;4.Based on the analysis methods and platforms,a microbiome analysis pipeline is integrated and published on public access data platform.Part 2 In Guangdong,we pre-selected 14 districts/counties.In each district/county,we further selected three neighborhoods/towns and in each neighborhood/town,we selected two communities/villages by using PPS(probability proportional to size)methods.We collected feces from all pariticipants and characterized fecal microbiome by sequencing the V4 variable reigons of 16S rRNA gene.We also collected other metadata for analysis.PERMANOVA is used to calculate the effect size of the metadata on gut microbiome varience.We used multivariate association analysis(MaAsLin)to determine significant associations between gut microbial components with host metadata that related to metabolic syndrome.Results:1.Optimization of analysis methods:When comparing sequences generated from different PCR primers but the same variable region,we found determining biomarkers using datasets that were generated from different technical parameters is higly affected.As all de novo clustering methods would generate unstable OTUs,which would further affect the calculation of rarefaction curves,beta-diversity and differential species calculations,we developed an Method based on references,which produce stable OTUs and reduce the process time by multi-thread.We bulked those methods into one analysis pipeline and published it on public accessible platform.2.The population-level gut microbiome survey:In the GGMP,8600 volunteers are included.And over 100 metadata are collected.The metadata and the gut microbiome are widely related.Among these metadata,geographical position has the largest effect size,which could be related to the local salt consumption behavious.For other metadata,age,Bristol stool score,body weight,urinal acid level,static time and diet have relatively larger effect on gut microbiome compared to the rest metadata.Almost all key biomarkers for metabolic syndrome that were reported in developed areas were reproduced in our cohort.On the other hand,gut microbiome in our population did show distinctions,e.g.,high gamma-Proteobacteria,which is negatively and significantly correlated with local economic development..Conclusions:1.We optimised microbiome analysis methods,regarding its stability and speed,and developed a pipeline from raw data to downstream analysis for microbiome research that is deposited in public platforms..2.Gut microbiome variation is widely correlated with host phenotypes,among which geographical distribution could be the most important contribution in dispersing gut microbiome composition.Moreover,gut microbiome in Eastern developing areas do show distinctions,e.g.,high abundance of Proteobacteria,that could increase the risk of metabolic diseases with lifestyle transition,indicating that more research on gut microbiome in develop areas can fulfill our knowledge in the relationship between gut microbiome and host health.