How to Use Fit-Hi-C R Package

Introduction

Fit-Hi-C is a tool for assigning statistical confidence estimates to intra-chromosomal contact maps produced by genome-wide genome architecture assays such as Hi-C. Compared to Python original, Fit-Hi-C R port has the following advantages:

Fit-Hi-C R package is more efficient than Python original by re-writting some logic in C/C++
Fit-Hi-C R package is easy to use for those familiar with R language without additional configuration

Install FitHiC

To install this package, start R and enter

## try http:// if https:// URLs are not supported
source("https://bioconductor.org/biocLite.R")
biocLite("FitHiC")

There are two ways to retrieve development versions

To get the latest development version, start R and enter

## try http:// if https:// URLs are not supported
source("https://bioconductor.org/biocLite.R")
biocLite("BiocInstaller")
useDevel()
biocLite("FitHiC")

To get a specified development version x.y.z, open a terminal and enter

wget http://bioconductor.org/packages/devel/bioc/src/contrib/FitHiC_x.y.z.tar.gz .
R CMD INSTALL FitHiC_x.y.z.tar.gz

Prepare Data

Before running Fit-Hi-C, two input files should be prepared.

FRAGSFILE This file stores the information about midpoints (or start indices) of the fragments. It should consist of 5 columns: first column stands for chromosome name; third column stands for the midPoint; fourth column stands for the hitCount; second column and fifth column can be arbitrary.

FRAGSFILE
Chromosome.Name	Mid.Point	Hit.Count	Column.5
1	1305	0	0
1	2635	233	1
1	4756	876	1
1	8568	1076	1
1	10384	1210	1
1	12246	639	1

INTERSFILE This file stores the information about interactions between fragment pairs. It should consist of 5 columns: first column and third column stand for the chromosome names of the fragment pair; second column and fourth column stand for midPoints of the fragment pair; fifth column stands for hitCount.

INTERSFILE
Chromosome1.Name	Mid.Point.1	Chromosome2.Name	Mid.Point.2	Hit.Count
10	100894	10	150593	2
10	100894	10	162267	1
10	100894	10	169783	2
10	100894	10	179515	3
10	100894	10	182528	1
10	100894	10	185071	1

Besides, OUTDIR, the path where the output files will be stored, is also required to be specified.

Run Fit-Hi-C

After the input data is well prepared, you can easily run Fit-Hi-C in R as:

library("FitHiC")
FitHiC(FRAGSFILE, INTERSFILE, OUTDIR, ...)

If you want to output images simultaneously, explicitly set visual to TRUE:

library("FitHiC")
FitHiC(FRAGSFILE, INTERSFILE, OUTDIR, ..., visual=TRUE)

Examples

The pre-processed Hi-C data is from Yeast - EcoRI ¹. FRAGSFILE and INTERSFILE are located in system.file("extdata", "fragmentLists/Duan_yeast_EcoRI.gz", package = "FitHiC") and system.file( "extdata", "contactCounts/Duan_yeast_EcoRI.gz", package = "FitHiC"), respectively. When input data is ready, run as follows:

library("FitHiC")
fragsfile <- system.file("extdata", "fragmentLists/Duan_yeast_EcoRI.gz",
    package = "FitHiC")
intersfile <- system.file("extdata", "contactCounts/Duan_yeast_EcoRI.gz",
    package = "FitHiC")
FitHiC(fragsfile, intersfile, getwd(), libname="Duan_yeast_EcoRI",
    distUpThres=250000, distLowThres=10000)

Internally, Fit-Hi-C will successively call generate_FragPairs, read_ICE_biases, read_All_Interactions, calculateing_Probabilities, fit_Spline methods. The execution of Fit-Hi-C will be successfully completed till the following log appears:

## Fit-Hi-C is processing ...
## Running generate_FragPairs method ...
## Complete generate_FragPairs method [OK]
## Running read_All_Interactions method ...
## Complete read_All_Interactions method [OK]
## Running calculating_Probabilities method ...
## Writing Duan_yeast_EcoRI.fithic_pass1.txt
## Complete calculating_Probabilities method [OK]
## Running fit_Spline method ...
## Writing p-values to file Duan_yeast_EcoRI.spline_pass1.significances.txt.gz
## Complete fit_Spline method [OK]
## Running calculating_Probabilities method ...
## Writing Duan_yeast_EcoRI.fithic_pass2.txt
## Complete calculating_Probabilities method [OK]
## Running fit_Spline method ...
## Writing p-values to file Duan_yeast_EcoRI.spline_pass2.significances.txt.gz
## Complete fit_Spline method [OK]
## Execution of Fit-Hi-C completed successfully. [DONE]
## .Primitive("return")

The output files come from two internal methods called by Fit-Hi-C.

calculate_Probabilites

Duan_yeast_EcoRI.fithic_pass1.txt
avgGenomicDist	contactProbability	standardError	noOfLocusPairs	totalOfContactCounts
10105	3.12e-05	2.7e-06	322	22212
10315	3.05e-05	2.5e-06	330	22251
10545	2.87e-05	2.1e-06	350	22191
10779	2.97e-05	3.0e-06	344	22583
10982	3.16e-05	2.7e-06	323	22532
11196	3.32e-05	2.7e-06	302	22185

Duan_yeast_EcoRI.fithic_pass2.txt
avgGenomicDist	contactProbability	standardError	noOfLocusPairs	totalOfContactCounts
10107	1.15e-05	8e-07	252	6428
10317	1.31e-05	9e-07	266	7709
10546	1.43e-05	8e-07	281	8887
10779	1.27e-05	8e-07	285	7974
10982	1.32e-05	8e-07	255	7426
11196	1.40e-05	8e-07	238	7356

fit_Spline

Duan_yeast_EcoRI.spline_pass1.significances.txt.gz
chr1	fragmentMid1	chr2	fragmentMid2	contactCount	p_value	q_value
10	100894	10	150593	2	0.9988785	1
10	100894	10	162267	1	0.9985433	1
10	100894	10	169783	2	0.9708609	1
10	100894	10	179515	3	0.8072602	1
10	100894	10	182528	1	0.9831568	1
10	100894	10	185071	1	0.9795001	1

Duan_yeast_EcoRI.spline_pass2.significances.txt.gz
chr1	fragmentMid1	chr2	fragmentMid2	contactCount	p_value	q_value
10	100894	10	150593	2	0.9813195	1
10	100894	10	162267	1	0.9902851	1
10	100894	10	169783	2	0.8983241	1
10	100894	10	179515	3	0.6547083	1
10	100894	10	182528	1	0.9571117	1
10	100894	10	185071	1	0.9501637	1

If visual is set to TRUE, corresponding images will be also outputed:

Support

For questions about the use of Fit-Hi-C method, to request pre-processed Hi-C data or additional features and scripts, or to report bugs and provide feedback please e-mail Ferhat Ay.

Ferhat Ay <ferhatay at uw period edu>