File Description

This repository is used for depositing the analysis scripts, derived data/results and corresponding figures and tables for the paper "PhyloMed: a phylogeny-based test of mediation effect in microbiome".

Here are brief descriptions for each folder, detailed explanations are commented within code and described in the paper:

• a folder named Analysis, which contains the real data analysis R scripts.

  • 0.prepareData.R
  • 1.runModel.R
  • 2.generatePlot.R
  • 3.runPairLeafAnalysis.R
  • 4.sensitivityAnalysis.R
  • 5.generateTab.R
  • *.utility.R

• a folder named Data, which contains raw data and derived data.

• a folder named Figs, which contains the figures in the main text and supplementary information.

• a folder named Tabs, which contains the tables in the main text and supplementary information.

• a folder named Simulation, which contains the simulation shell scripts and R scripts.

  • ProposedModel_Main: The mediating OTUs are clustered on the tree. The tree is constructed from top 100 OTUs.
  • RareTaxa: The mediating OTUs are clustered on the tree. The tree is constructed from all OTUs.
  • IncreaseNumOfCausalTaxa: The mediating OTUs are randomly scattered with large or small effect size.
  • DiffModels: Evaluate empirical type I error of global mediation test when different subcomposition models were used.
  • PropOfNull: Evaluate bias and standard deviation of the estimated probabilities of three sub-nulls using different estimation approaches.
  • PseudoCount: Evaluate empirical type I error and power of global mediation test when different pseudo counts were used.
  • IncreaseEffectSize: Evaluate empirical FDR when the mediating OTUs are clustered on the tree with relatively small effect size.
  • Download a copy of R (R413.tar.gz) and create a portable copy of necessary R packages (See tutorial) OR you can obtain the R413.tar.gz and packages_med.tar.gz from Zenodo. By submitting the ./medtest.dag to High-Throughput Condor, you are expected to get a csv file in the corresponding RESULT folder.
  • generateData.R: Generate the zeeviD_*.Rdata, which is used as a basis in the simulation.
  • generateTxt.R: Generate the input.txt and comb_sim.txt in GETSIM and COMBSIM folder, respectively.
  • zeeviD_*.Rdata: The phyloseq-object used in the simulation.

R Package

The proposed method PhyloMed in the paper is implemented in the R package miMediation, which is publicly available at https://github.com/KiRinHong/miMediation

Contact

• Qilin (Kirin) Hong - [email protected]