|
|
[求助]
請教關于effect size analyses的內(nèi)容
小弟非生物統(tǒng)計專業(yè)背景，統(tǒng)計背景非常薄弱，現(xiàn)在研究方向是微生物組學，看到某大牛文章中涉及到關于Effect size analyses的內(nèi)容，實在是看起來費勁，不太懂這里邊的原理，更別說具體操作了，還請大神能詳細指導一下(最好能詳細的介紹一下這里邊的原理和具體如何操作，需要什么樣的數(shù)據(jù)等等)，謝謝。
原文材料與方法內(nèi)容如下：
Effect size analyses. In this study, we integrated the multi-omics analyses method which could parse the relationship of metabolome, gut microbiome and phenome. We performed the“effect size” analysis strategy to determine whether the omic datasets can affect each other. To assess the proportion of variance of an omic dataset (i.e. serum metabolome) that be explained by another omic dataset (i.e. phenome), firstly, the adonis function of the R package vegan was used to estimate the “one-to-all” effect size (R2) between each single variable of the secondary omic (phenome) to the whole original omic dataset (serum metabolome). Only variables with significant (P < 0.05, 999 permutations) effect on the original omic dataset were considered later. Then, to get rid of redundant variables, the Pearson correlation coefficient between variables was calculated, and variables with correlation coefficient greater than 0.5 were removed. Finally, the combined effect size was calculated based on all non-redundant variables using adonis function.
The relationships of all variables among omic datasets were established using correlation network analysis as follows: 1) Spearman correlation coefficient between serum metabolome, faecal metabolome and gut microbiome with correlation coefficient (rho) greater than 0.35 was calculated, the P value was determined and the threshold of 0.01 was accepted; 2) to identify which are the key substances in the network, the entire network was parsed, the number of connections of every serum metabolome cluster was calculated. The correlation relationships of omic variables were calculated on both ESRD patients and healthy controls, respectively. Finally, the network diagrams were visualized by Cytoscape using circular layout.
Comparing of effect size of gut microbiota on the metabolomes in different studies. Four different studies were involved, including the European diabetes study (371 samples), the Chinese obesity study (151 samples), the Chinese ACVD study (102 samples), and this study (292 samples). The pre-processing methods for these raw data were the same as our study. To obtain comparable results on their taxonomic composition, high quality sequences of four studies were mapped onto the integrated gut microbial gene catalogue (IGC), which had been constructed by the yet largest number of human gut metagenome samples. After the previous analysis of taxonomic assignment, only species-level taxonomic profiles of these four studies were used to analyze the effect size of gut microbiota on the host serum and faecal metabolomes. The method of effect size calculating was described above. |
|
回復此樓
生物信息學或者生物相關專業(yè) （學碩）
已經(jīng)有0人回復
6+6