MrTADFinder aims to identify topologically associating domains (TADs) in multiple resolutions.

MrTADFinder takes an whole-genome-to-whole genome contact map as an input. The contact map should be a sparse matrix stored in a tab-delimited file as follow

1 1464 39
1 1768 19
1 4455 20
1 4458 70
1 7368 29
1 10413 25
1 10563 30
1 10687 37
1 10690 24
1 11123 28
...

The first 2 columns are indices of genomic bins, and the third column represents the contact frequency (corresponding matrix elements). This simple format has been widely used by mapping tools such as HiCPro.

The mapping between chromosome and genomics bins is specified by two annotation files.

To bin the human genome (hg19) in 40kb bin size, an annotation file in the following format is provided (see ./data/bins_file1):

1 chr1 0 6231
2 chr2 6232 12311
3 chr3 12312 17262
4 chr4 17263 22041
5 chr5 22042 26564
6 chr6 26565 30842
7 chr7 30843 34821
8 chr8 34822 38481
9 chr9 38482 42012
10 chr10 42013 45401
11 chr11 45402 48777
12 chr12 48778 52124
13 chr13 52125 55004
14 chr14 55005 57688
15 chr15 57689 60252
16 chr16 60253 62511
17 chr17 62512 64541
18 chr18 64542 66493
19 chr19 66494 67972
20 chr20 67973 69548
21 chr21 69549 70752
22 chr22 70753 72035
23 chrX 72036 75917
24 chrY 75918 77402
25 chrM 77403 77403

The first and second columns show the indices and names of various chromosome. Based on a bin size of 40kb, chromosome 1 is divided into 6232 bins (from bin 0 to bin 6231). Apart from the last bin, all the bins are 40kb. In this example, the whole human genome is divided int 77404 bins, and therefore the corresponding contact map is a square matrix of size 77404. Numbers running from 1 to 77404 are used to indexing the contact map file.

The 2nd file has the form

0 1 40000
0 40001 80000
0 80001 120000
0 120001 160000
0 160001 200000
0 200001 240000
0 240001 280000
0 280001 320000
0 320001 360000
0 360001 400000

IN this case, there are 77404 lines, representing all genomic bins (chromosome number (0=chr1 etc), start and end points).

Annotation files based on binning human genome (hg19) in 40kb are provided. Users can use different annotation files for different bin sizes or different organisms.

Output file is simply a csv file that stores a list of TADs (chromosome number, start and end coordinates).

"chr","domain_st","domain_ed","domain_st_bin","domain_ed_bin","idx"
"chr10",40001,200000,42015,42018,1
"chr10",200001,880000,42019,42035,2
"chr10",880001,1120000,42036,42041,3
"chr10",1120001,1160000,42042,42042,4
"chr10",1160001,1200000,42043,42043,5
"chr10",1200001,1240000,42044,42044,6
"chr10",1240001,1280000,42045,42045,7
"chr10",1280001,1320000,42046,42046,8
"chr10",1320001,1360000,42047,42047,9

The bins defined by the annotation files are provided in the 4th and 5th columns.

MrTADFinder is written in Julia. It has been tested in Julia v0.4.3. If Julia and the required packages are installed (see the first few lines in MrTADFinder.jl), one could simply run in the command prompt

julia run_MrTADFinder.jl contact_map ./data/bins_file1 ./data/bins_file2 res=1.0 10 TAD_chr10.bed

The 1st agrument: contact map.
The 2nd and 3rd agruments are the 2 annotation files.
The 4th argument is the resolution parameter.
The 5th argument is the chromosome of interest.
The 6th argument is the path and name of the TAD output file. An additional file of boundary score will be generated.