A/B compartment analysis

This section allows to calculate the principal components (PCs) of the Pearson correlation matrix that can be used to delineate A/B compartments in Hi-C data at low resolution (usually 1 Mb or 500 kb). It is possible to calculate both PC1 and PC2. Usually, the sign of the eigenvector (PC1) indicates the compartment.

Note that comparing PC values (either PC1 or PC2) between experiments may not be appropriate. While PC1 usually correlates with active and inactive compartments, the nature of this association (i.e. which PC sign (+,-) is associated to active or inactive regions) may be different among experiments. Instead, it is always recommended to compare interaction profiles of a specific genomic locus towards the other loci.

In order to directly compare eigenvectors among two experiments, first you should make sure that the association between PC1 and active and inactive compartments is the same for both the experiments. Usually, "A compartments" (positive PC values) are considered associated to gene rich and active genomic regions, while "B compartments" (negative PC values) to gene poor and inactive regions. In order for the sign of the eigenvalues to match our hypothesis, you could calculate the correlation of the eigenvector (separately per each chromosome) with a gene density track at the same resolution (1 Mb for example). If the correlation is positive, the eigenvector is left as it was, otherwise the sign of its values should be flipped.

Table of contents

1. Calculating the principal component
2. Plotting the principal component

1. Calculating the principal component

HiCtool allows to calculate either the first (typically used) or the second principal component of the Person correlation matrix. In order to do so, the Person correlation matrix has to be calculated first as presented here.

This is the code to calculate the first principal component:

python2.7 /HiCtool-master/scripts/HiCtool_compartment_analysis.py \
--action calculate_pc \
-c /HiCtool-master/scripts/chromSizes/ \
-b 1000000 \
-s hg38 \
--chr 6 \
--pc PC1

where:

• --action: Action to perform (here calculate_pc).
• -c: Path to the folder chromSizes with trailing slash at the end /.
• -b: The bin size (resolution) for the analysis.
• -s: Species name.
• --chr: The chromosome to be used.
• --pc: Which principal component to be returned (either PC1 or PC2).
• --flip: Set to -1 if you wish to flip PC values.

The output is a txt file with the values of the principal component selected, in this case chr6_1000000_PC1.txt inside the folder ./yaffe_tanay_1000000/.

The parameter --chr can be used also to pass multiple chromosomes (as a list between square brackets) at once and also multi-processing computation is provided if your machine supports it, using the parameters -p followed by the number of processors to be used in parallel (for example -p 2):

chromosomes=[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,X,Y]

python2.7 /HiCtool-master/scripts/HiCtool_compartment_analysis.py \
--action calculate_pc \
-c /HiCtool-master/scripts/chromSizes/ \
-b 1000000 \
-s hg38 \
--chr $chromosomes \
--pc PC1

2. Plotting the principal component

The following code allows to plot the principal component values in a barplot, which allows to visually delineate compartments. The following plot usually is presented together with the Pearson correlation matrix.

python2.7 /HiCtool-master/scripts/HiCtool_compartment_analysis.py \
--action plot_pc \
-i ./yaffe_tanay_1000000/chr6_1000000_PC1.txt \
-c /HiCtool-master/scripts/chromSizes/ \
-b 1000000 \
-s hg38 \
--chr 6 \
--pc PC1 \
--plot_grid 0 \
--plot_axis 1

where:

• --action: Action to perform (here plot_pc).
• -i: Input file with the principal component values calculated above.
• -c: Path to the folder chromSizes with trailing slash at the end /.
• -b: The bin size (resolution) for the analysis.
• -s: Species name.
• --chr: The chromosome to be used.
• --pc: Which principal component to be returned (the same you have calculated above).
• --plot_grid: Set to 1 if you wish to plot the grid.
• --plot_axis: Set to 0 or remove this parameter if you do not wish to plot the axis. This may be useful to obtain a simpler plot that you may want to visualize together with the correlation matrix.