.. _DIFF:

Capture Comparative Analysis
============================

Introduction
------------

`Chicdiff <https://academic.oup.com/bioinformatics/article/35/22/4764/5514042>`_ is an R package complementary to CHiCAGO, with the goal of identifying differential Capture Hi-C interactions. This can be very useful for comparing interactions found under different experimental conditions. 


For detailed information on the package and installation, please see `Chicdiff Github page <https://github.com/RegulatoryGenomicsGroup/chicdiff>`_. Here is a brief step by step installation (requires R version >= 3.4.0)

.. code-block:: r

   install.packages("devtools")
   library(devtools)
   install_github("RegulatoryGenomicsGroup/chicdiff", subdir="Chicdiff")

Input files
-----------

Peak files
++++++++++

Chicdiff uses a peak format of the interactions, based on the :ref:`CHiCAGO output<CHIOUT>` 

**Generating peak files:**

First, generate a tab delimited list of all the samples and their associate Rds files (as generated under `data` directory from a :ref:`CHiCAGO run <CHIRUN>`), for which you want to find differential interactions. Make sure that all the datasets that you use were generated with the same :ref:`Design files<DESDIR>`.

For this demonstration we will compare interactions between NSC (two replicas) and iPSC (two replicas).  This information should be recorded in a tab delmited file, as in this example file, named *Rd_calls.txt* with the following content:


*Rd_calls.txt:*

.. code-block:: text

   NSC_rep1     /home/user/NSC_rep1_chicago_calls/data/NSC_rep1_chicago_calls.Rds
   NSC_rep2     /home/user/NSC_rep2_chicago_calls/data/NSC_rep2_chicago_calls.Rds
   iPSC_rep1    /home/user/iPSC_rep1_chicago_calls/data/iPSC_rep1_chicago_calls.Rds
   iPSC_rep2    /home/user/iPSC_rep2_chicago_calls/data/iPSC_rep2_chicago_calls.Rds


The makePeakMatrix.R script generate a peak database based on the CHiCAGO databases that were detailed in the Rd_calls.txt file.

.. code-block:: console

   Rscript ./chicago/chicagoTools/makePeakMatrix.R --twopass --notrans --maxdist 3000000 Rd_calls.txt all_peaks

In this example peaks of cis interactions <3Mb are called.


If multiple (>2) datasets were used in the makePeakMatrix.R step, a clustered dendrogram will be generated as well:

.. image:: images/dendro1.png

To analyze the data you will also need:

Design Library
++++++++++++++

Path to design library (same one as was used for generating the .Rds files above) 
  
Chinputs
++++++++

chinput files (same as the chinput files that were used for the CHiCAGO runs)


Calling differential interactions
---------------------------------

In your R console, follow the principle of this example, with two replicas of NSCs and two replicas of iPSCs, make sure to be consistent with naming (e.g. to replica no. 1 of iPSC experiment use the same name for example *iPSC_rep1* for all its associated files such as chinput, Rds etc').

.. code-block:: r

   library(Chicdiff)

   #set design library:

   DesignDir <- file.path("/home/user/", "h_10kDesingFiles")

   #set interaction data sets to be included:

   ChicagoData <- list(iPSC_Micro = c(iPSC_rep1 = file.path("/home/user/10kb_run/iPSC_rep1/data","iPSC_rep1.Rds"), iPSC_rep2 = file.path("/home/user/10kb_run/iPSC_rep2/data","iPSC_rep2.Rds")), NSC_MicroC = c(NSC_rep1 = file.path("/home/user/10kb_run/NSC_rep1/data","NSC_rep1.Rds"), NSC_rep2 =  file.path("/home/user/10kb_run/NSC_rep2/data","NSC_rep2.Rds")))
   
   #set chinput files:

   CountData <- list(iPSC_MicroC = c(iPSC_rep1 = file.path("/home/user/pooled_10kb_chinput/iPSC_rep1", "iPSC_rep1.chinput"), iPSC_rep2 = file.path("/home/user/pooled_10kb_chinput/iPSC_rep2", "iPSC_rep2.chinput")), NSC_MicroC = c(NSC_rep1 = file.path("/home/user/pooled_10kb_chinput/NSC_rep1", "NSC_rep1.chinput"), NSC_rep2 = file.path("/home/user/pooled_10kb_chinput/NSC_rep2", "NSC_rep2.chinput")))

   #set peakfiles:

   PeakFiles <- "/home/user/peaks/all_peaks.txt"

   #set parameters for Chicdiff run:

   chicdiff.settings <- setChicdiffExperiment(designDir = DesignDir, chicagoData = ChicagoData, countData = CountData, peakfiles = PeakFiles, outprefix="10kb", settings = list(parallel=TRUE))

   #run Chicdiff

   Diff <- chicdiffPipeline(chicdiff.settings)

   #example, how to plot specific bait interactions using the bait number:

   plotDiffBaits()



.. image:: images/diff_DCAF13.png
   :width: 300 px
   :align: center

Additional information on how to use the package can be found on `Chicdiff Vignette <http://functionalgenecontrol.group/wp-content/uploads/Chicdiff.html>`_