Skip to content

Latest commit

 

History

History
94 lines (69 loc) · 4.86 KB

README.md

File metadata and controls

94 lines (69 loc) · 4.86 KB
Raw

Linkage Analysis in R

Friendly Reminder

If you use or take inspiration from this repository please cite with this link: santurini/DAG-Linkage-Analysis-in-R

Your support will be truly appreciated and feel free to contact me at my following links or just send me an email:

Overview

In this work we implemented to universal hypothesis test to check whether a link or a path existed or not in a directed acyclic graph. First we checked the properties of the tests and then we applied both on a real case study.

Content

scripts this folder contains the scripts:

  • functions.R which contains all the functions to run the scripts
  • LinkageTest.R which contains the implementation of the test for a single link
  • PathwayTest.R which contains the implementation of the test for a directed pathway
  • Cytometry.R which contains the application of the tests on real data
  • plotSparsity.R which contains the script fro the scatter plots of the test results with different sparsity values

report this folder contains the final report of the case study:

  • Report.html is the knitted html file written in rmarkdown

data is a folder which contains:

  • cytometry-data.xlsx the excel file with 9 different sheets

Theory concepts

The aim of this repository is to implement the Universal Hypothesis Test that is explained in the following image:

Test of graph linkages

This is the definition we used to implement the graph linkage test:

Test of directed pathway

This is the definition we used to implement the graph pathway test:

Tools

To be able to implement it we used the clrdag package which contains the MLEdag function, the linck to the repository is the following:

Data

Once implemented we tested them on real world data regarding Cell Signaling represented as a DAG:

Overview

  1. Estimating the variance and log-likelihood:

    • Sigma estimate:

    • log-Likelihood estimate:

  2. Implementing the LRT (Likelihood ratio test) and Crossfit test

  3. Implementing the Universal tests on random data to compute size and power

    • Formula to generate the random data:

  4. Applying the test on real world data to check linkages between proteins