The Methods Of Constructing Soil Mircrobial Ecological Network And Its Comparison

For a long time, the vast majority of research on soil microbial communities address the questions as to what microorganisms and how many microorganisms are there in the soil. Understandings on roles of microorganisms have been largely achieved by correlating microbial lineages, community diversity and structure with soil variables. Little attention has been paid to the interaction between microorganisms. But, in fact, the microorganisms in soil rarely live in isolations. Rather, they always experience interactions with other microorganisms, such as symbiosis, parasitism, alternate, antagonism, and so on. Traditionally, the interactions between microorganisms are determined by co-culture experiments. The most typical example is to determine the antagonism:cultivate microbe A and B in the same plate. If an inhibition zone between the parties, it means there is the antagonistic action between A and B.Although co-culture is still a standard method to determine the interaction between microorganisms, but it has obvious limitations to study the interactions in the microbial community. First, it only applies to those cultivable microbes, but the vast majority of microbes in the soil is uncultivable. Second, the conditions of soil is very different from the plate, the interactions appeared in the plate may not be the interactions practically occurring in the soil. Therefore, we must find other ways to study the interactions in the microbial community.In recent years, researchers have developed a lot of softwares for studying the interactions between members of the microbial community. These softwares can utilize the results of high-throughput sequencing to predict the interaction between microbes. High-throughput sequencing reflects the in-situ situation of soil microbial community. Compared with the plate, data derived from high-throughput sequencing are more closely reflecting microorganisms in situ soil conditions. These software make it possible for the study of microbial interactions between community level. However, the parameters of the software are usually different from user to user. There has been no report showing how to choose the parameters and the comparison of different results caused by different parameters. That is to say, choose different parameters may get different results, but it is unknown that what kind of result is more reasonable.Compared with the protein interaction relations, the interaction relationship between soil microorganisms still have no reference database, so the network have no right or incorrect predictions. However, a group of good parameters should be able to get the prudent result. If a particular interaction between a pair of microorganisms is objective existence, it can be detected at different OTU cutoff level. Therefore, different OTU cutoff level becomes a reference baseline testing network robustness. Based on this hypothesis, this paper discussed how to choose the parameters.This article compared two network analysis methods, CoNet and MENA, discussed how to choose appropriate parameters to obtain robust results. First, we introduced a set of parameters and the procedures of two software. Then, high-throughput sequencing data made up by 16 soil samples were prepared. By setting up different parameters in combinations, networks were produced by two software respectively, and compared the difference between the results. Finally, the reproducibility of the modules partitioned from the network constructed by CoNet and MENA were evaluated. Module partitioning methods, one that is embedded in MENA and the one by OH-PIN algorithm (in default settings) were also compared. Based on the comparisons on the reproducibility of the modules at different OTU cutoff level, a set of parameters were recommended. The results shows that using the recommended parameters combination of CoNet can get microbiological ecological network with strong robustness. The article provides two methods of soil microbial ecological network analysis, and recommends a set of software parameters.