Welcome to the deepBlockAlign web page
[Download]
- deepBlockAlign v1.3.1 - updated readme file with descriptive text on input dataset requirement (Dec 04, 2014).
- deepBlockAlign v1.3 - bug fix that was leading to out of memory error for alignment of read profiles containing block of >500 nt length or number of blocks >100 (October 15, 2014)
- deepBlockAlign v1.2 - added new parameters for visualizing alignments (-f) and allowing alignment of sequenctial pairs from two input files (-p) (August 15, 2013)
- deepBlockAlign v1.1 - no rounding of normalized read count. Makes the alignment score more accurate (December 14, 2012)
- deepBlockAlign v1.0 - release (June 27, 2011).
[Programs and datasets]
The directory 'deepBlockAlign_v*' contains following programs and datasets
1. deepBlockAlign.x: program to align two or more RNA-seq read block profiles 2. deepBlockAlign.c: source code of deepBlockAlign 3. plotdeepBlockAlign.pl: wrapper program to deepBlockAlign. Also output alignments in graphical format 4. plotblockAlign.r: accessory R program to plotdeepBlockAlign.pl 5. test_suite/human_eb.blockGroups: Benchmark dataset used in current study
[Installation]
To install deepBlockAlign, download deepBlockAlign.tar.gz and unpack it
tar -zxvf deepBlockAlign.tar.gz A directory, deepBlockAlign will be created, now execute
make or make all which compiles deepBlockAlign and creates an executable
deepBlockAlign.x
[Usage]
deepBlockAlign.x is called with the following parameters
deepBlockAlign.x -q <query file> -s <subject file> [ARGUMENTS]
Available options are:
========================================================== DeepBlockAlign a tool for aligning RNA-seq profiles of read block patterns ========================================================== usage: deepBlockAlign.x -q <file> -s <file> [ARGUMENTS] [INPUT] -q <file> query file (blockbuster output) -s <file> subject file (blockbuster output) [ARGUMENTS: Block Group Alignment] -d <double> distance weight (default: 6.0) -b <double> block similarity weight (default: 1.0) -g <double> gap penalty (default: -1.0) -a <int> print alignment 1=on (default: 0) -l <file> create file containing alignment score of all blocks against all blocks -f <file> create file necessary to plot block group alignment -p <int> only align pairs of block groups e.g 1 with 1, 2 with 2 etc (default:0) [ARGUMENTS: Block Alignment] -c <int> gap initialization penalty (default: -2) -e <int> gap extension penalty for (default: -1) -t <double> threshold for expression difference between two blocks (default: 1.0) -m <int> match score (default: 1) -n <int> mismatch score (default: -1) -z <int> profile shape difference penalty (default: 1) -h this helpful message [VERSION] v1.3 (Oct 15, 2014)
plotdeepBlockAlign.pl is called with the following parameters.
plotdeepBlockAlign.pl -i <query file> -j <subject file>
Available options are:
Program: plotdeepBlockAlign.pl (align and visualize the alignment of read profiles) Author: University of Copenhagen, Denmark Version: 1.0 Contact: sachin@rth.dk Usage: plotdeepBlockAlign.pl -i <file> -j <file> [OPTIONS] -i <file> [first block group file] -j <file> [second block group file] -f <id> [first block group id] -s <id> [second block group id] -p <dir> [directory with programs (deepblockalign.x etc)] -o <dir> [directory with output files] -x <string> [prefix for temporary files, useful when running on queue] -l [output in HTML format] -a <string> [path to hyperlink alignment images from the HTML format] -q [output in plain text format] -k [keep the block and block group alignment files] -v <string> [first block group title (default: first)] -w <string> [second block group title (default: second)] [OPTIONS: Block Group Alignment] -d <double> [distance weight (default: 1.0)] -b <double> [block similarity weight (default: 1.0)] -g <double> [gap penalty (default: -1.0)] [OPTIONS: Block Alignment] -c <int> [gap initialization penalty (default: -2)] -e <int> [gap extension penalty (default: -1)] -t <double> [threshold for expression difference between two blocks (default: 1.0)] -m <int> [match score (default: 1)] -n <int> [mismatch score (default: -1)] -z <int> [profile shape difference penalty (default: 1)]
[Example]
An usage example of deepBlockAlign is shown below. Example Block groups shown in the manuscript are provided with the downloads along with complete benchmark dataset.
$./deepBlockAlign.x -q testsuite/snoRNA-HACA-E3.example -s testsuite/hsa-mir-9-1.example # deepBlockAlign.x v1.3 started at Thu Dec 4 16:25:12 2014 # query: testsuite/snoRNA-HACA-E3.example # subject: testsuite/hsa-mir-9-1.example # parameters (block group alignment) # distance weight: 1.0; block similarity weight: 1.0; gap penalty: -1.0 # parameters (block alignment) # match: 1; mismatch: -1; threshold: 1; shape difference penalty: 1 # gap penalty (initialization): -2; gap penalty (extension): -1 >cluster_929|E3|chr3:187987781-187987842(+) >cluster_30|hsa-mir-9-1|chr1:154656768-154656831(-) 0.554014
$./plotdeepBlockAlign.pl -i testsuite/human_eb.blockGroups -j testsuite/human_eb.blockGroups \ -f cluster_176 -s cluster_531 -p . # plotdeepBlockAlign started at Thu, 04 Dec 2014 16:27:37 +0100 # first block group: cluster_176 # first file: testsuite/human_eb.blockGroups # second block group: cluster_531 # second file: testsuite/human_eb.blockGroups computing alignment....done (score: 0.72) aligning blocks...done aligning block groups...lsdone (cluster_176_n-a--cluster_531_hsa-mir-424.pdf) creating final output file... done (./cluster_176_n-a--cluster_531_hsa-mir-424_FINAL.pdf)
[Input Data Format]
deepBlockAlign takes the input data in the blockbuster output format. blockbuster is a program to define block groups based on the mapping pattern of short reads. A typical input comprises
a) Header section (11 tab delimited fields)
>cluster_929 chr3 187987781 187987842 + 306.50 194 2 E3 snoRNA Intron
b) Reads section (7 tab delimited fields)
chr3 187987783 187987809 morin_EB_685629 0.500000 + 1 chr3 187987783 187987806 morin_EB_685761 0.500000 + 1 . . . chr3 187987783 187987806 morin_EB_685630 0.500000 + 1
Important Note: The header section from blockbuster consists of eight fields (id, chr, start, end, strand, #reads, #tags, #blocks). Before providing it as an input to deepBlockAlign, three additional fields (name, annotation, locus) needs to be added. The information in these fields is required for most downstream analysis. If not available, user can put a "." in place of these fields.
a) Header section (8 tab delimited fields)
>cluster_929 chr3 187987781 187987842 + 306.50 194 2
[Details]
For details, please refer to 1. deepBlockAlign publication 2. blockbuster publication