Research in Genomics

Eric Rivals
LIRMM (Computer Science Department)
CNRS - Université de Montpellier
France
rivals_AT_lirmm.fr
http://www.lirmm.fr/~rivals

1 Supplementary material for "Combining SAGE tags to predict genomic transcribed regions"

1 Tandem SAGE tags approach for transcriptome annotation

In a project in collaboration between the Laboratoire d'Informatique de Robotique de Microélectronique de Montpellier, the Institut de Génétique Humaine (I.G.H.), and the Helsinki University of Technology, we propose a new method to annotate transcriptionnally active regions on any mammalian genome. For this we developped with the team of J. Tarhio (H.U.T.) a program able to search for pairs of SAGE or MPSS tags in tandem on the genome. A tag anchored by Sau3A1 is called a G-tag, while a tag anchored by NlaIII is termed a C-tag.

Two programs named SearchTandemSAGE-CG and SearchTandemSAGE-GC respectively are available as a Linux executable (32 byte architecture) and as Mac OS X executable for download free of charge for academic users.

SearchTandemSAGE-CG: : searches for C-G type pairs where for a given G-tag, the associated C-tag is the nearest 5' C-tag that belongs to the input list of C-tags
SearchTandemSAGE-GC: : performs the dual search, that is to say it searches for G-C type pairs where for a given C-tag, the associated G-tag is the nearest 5' G-tag that belongs to the input list of G-tags.

The usage are similar for both programs and we detail it for only one of them.

1.1 Usage

Command (Linux/Unix):
```
SearchTandemSAGE-CG -pc pattern-file-Ctype -pg pattern-file-Gtype -d dna-sequence-FASTA
```
or with output redirection:
```
SearchTandemSAGE-CG -pc pattern-file-Ctype -pg pattern-file-Gtype -d dna-sequence-FASTA > output-file
```
The program searches for pair of adjacent patterns along the sequence such that each reported pair contains C-type pattern (leftmost pattern in the pair or in biological terms at its 5' end) then one G-type pattern (rightmost pattern in the pair or in biological terms at its 3' end). To be reported as a match the substring must match exactly the corresponding patterns. There is no notion of approximate matching in this framework.

To search for pairs of the form (G-tag, C-tag), use SearchTandemSAGE-GC in a similar fashion.
Input: the program takes as input two files containing patterns of different types, and one dna sequence in FASTA format.
Pattern file format: simple text file with one pattern per line. The pattern should be complete, e.g., for SAGE tags it should include the restriction site (either CATG or GATC) at its beginning.
Output: it is printed on the standard output. For convenience, it should be redirected in an output file as shown above (Usage with output redirection). The output is in simple text format, organised in 8 columns, which are separated by semicolumns. This is also known as CSV format, and can be loaded with any spreadsheet program (Excel, OpenOffice Calc, etc.). A sample output is shown below:
```
>test1
210498;C;catgtgtacataca;54;115;G;gatccgcccgcctc;74;
210498;C;catgtgtacataca;54;23222;G;gatccaaatattgt;94;
230447;C;catgtggtattaga;108;185;G;gatccacctgcctc;120;
```
The first line gives the sequence ID taken from the sequence input file in FASTA format. Each of the following line reports the combined match of a pair of patterns. The line contains 8 columns, which are split in two groups of 4 columns organised similarly. Each group of 4 describes each pattern match in order of appearance in the sequence.

Column number 1 2 3 4

Meaning the pattern's Number Type (C or G) Sequence Position in the sequence

For one execution of SearchTandemSAGE searches for one type of pairs: a C-type pattern followed by a G-type pattern. Thus, the patterns describes on the first four columns are always of the same type.

The sequence positions are number from 0 on. Thus, a pattern occurring at the first sequence position is output with an occurrence position of 0.

Only positions of the sequence bearing a symbol in "acgtACGT" can match a position in an input pattern. Other symbols imply a mismatch in any comparison.

1.2 Examples

Input files:

C-tags file: tagsC.txt
G-tags file: tagsG.txt
sequence file: dna.fa

Commands:

Usage / Help

> ./SearchTandemSAGE-CG -h
Usage: ./SearchTandemSAGE-CG -h  print this helps
Usage: ./SearchTandemSAGE-CG -pc <pattern-file-Ctype> -pg <pattern-file-Gtype> -d <sequence-file-FASTA>
Default: <pattern-file-Ctype> tagsC.txt
Default: <pattern-file-Gtype> tagsG.txt
Default: <sequence-file-FASTA> dna.fa

SearchTandemSAGE-GC -h
Usage: SearchTandemSAGE-GC -h    print this helps
Usage: SearchTandemSAGE-GC -pc <pattern-file-Ctype> -pg <pattern-file-Gtype> -d <sequence-file-FASTA>
Default: <pattern-file-Ctype> tagsC.txt
Default: <pattern-file-Gtype> tagsG.txt
Default: <sequence-file-FASTA> dna.fa

Search C-G pairs without redirection

./SearchTandemSAGE-CG -pc tagsC.txt -pg tagsG.txt -d dna.fa 
>testdna
2;C;catgctatttagtt;16;3;G;gatcagggctgagg;30;
5;C;catgtcagtttgga;46;4;G;gatcttctacttgc;60;
6;C;catgtgggcgcctt;76;1;G;gatcctcagcctcc;90;
3;C;catgcgctgtgtgc;106;2;G;gatcaaataaaaaa;120;

Search C-G pairs with redirection

./SearchTandemSAGE-CG -pc tagsC.txt -pg tagsG.txt -d dna.fa > dna.resCG

Output file: dna.resCG

Search for G-C pairs with redirection

./SearchTandemSAGE-GC -pc tagsC.txt -pg tagsG.txt -d dna.fa > dna.resGC

Output file: dna.resGC

1.3 Download

Linux 32 bytes: : executable for SearchTandemSAGE-CG SearchTandemSAGE-GC
Mac OSX (> 10.3): : executable for SearchTandemSAGE-CG SearchTandemSAGE-GC

This document was translated from L^AT_EX by H^EV^EA.

Column number	1	2	3	4
Meaning the pattern's	Number	Type (C or G)	Sequence	Position in the sequence