TransGeneScan
=============
TransGeneScan is a gene finding tool for Metatranscriptomic sequences. TransGeneScan incorporates strand-specic hidden states, representing coding sequences in sense and anti-sense strands on transcripts in a Hidden Markov Model similar to the one used in FragGeneScan (https://github.com/COL-IU/FragGeneScan), and can predict a sense transcript containing one or multiple genes (in an operon) or an antisense transcript.

Installation
=============
To install TransGeneScan, please follow the steps below:

1. Clone this repository:
	git clone https://github.com/COL-IU/TransGeneScan.git

2. Make sure that you also have a C compiler such as "gcc" and "perl" and "python" interpreter.

3. Run "makefile" to compile and build excutable
	make tgs


Running the program
====================
1.  To run TransGeneScan, 

./run_TransGeneScan.pl -in=[seq_file_name] -out=[output_file_name] (-thread=[number of thread; default 1])

[seq_file_name]:		sequence file name including the full path
[output_file_name]:	output file name including the full path
[num_thread]:				number of thread used in TransGeneScan (optional parameter). Default 1.


Assembly of Transcripts
=======================
1. To assemble transcripts based on read mappings onto a single reference genome,

./scripts/pipeline.sh [reference_file] [reads_prefix] [TGSHome] [n] [k] [t]

[reference_file]: reference sequence file including full path
[reads_prefix]: paired-end reads prefix (not including _1.fastq, _2.fastq) including full path. The suffixes, _1.fastq and _2.fastq, are added within the script. Please make sure the files are named appropriately. 
[TGSHome]: Full path of TransGeneScan home directory
[n],[k],[t]: These are bwa parameters (please see bwa documentation for more information). The values used for testing were 4,4,4

Source files included
=====================
1. run_hmm.c, util_lib.c, util_lib.h, hmm.h, hmm_lib.c
These files contain the main Hidden Markov Model (HMM) framework of the prediction system. Most of the code is re-used from FragGeneScan as is. 

2. run_TransGeneScan.pl
This script is the main front end for the user to call the program for prediction. (See "Running the program" above)

3. post_process.pl
This script is part of the original FragGeneScan which makes corrections in the position of start codon based on a prediction model (See reference for more details). This code is re-used as is, in TransGeneScan.

4. FGS_gff.py
This script converts the TransGeneScan output format (which is the same as FragGeneScan output format) into gff format. 

5. processFragOut.py
This script is used to output predictions on sense transcripts and antisense transcripts as separate files.

6. train/*
These files include the training parameters used by the HMM. 

7. scripts/*
These scripts are used to do the assembly of transcripts based on read mappings (See "Assembly of Transcripts" above). 

Sample files included
=====================
1. example/transcripts.fasta
This file is the transcript assembly output produced by running scripts/pipeline.sh using paired-end reads downloaded from Short Reads Archive (SRR442380) mapped on to E.coli (NC_000913) as reference.

2. example/TGSout.out, TGSout.ffn, TGSout.faa, TGSout.gff
Prediction output from TransGeneScan in TGS output format (see below), nucleic acid fasta format, amino acid fasta format and gff format.

3. example/TGSout.sn
Prediction output from TransGeneScan containing only sense transcripts in TGS output format. 

4. example/TGSout.as
Prediction output from TransGeneScan containing only antisense transcripts. Since each entire transcript is an antisense transcript, no start/stop ranges are specified. The nucleotide sequence (same as input) is also output for these transcripts. 

5. example/TGSout.rem
List of headers from the input (transcripts.fasta), for which no predictions can be made. 

Number of assembled transcripts: 5999 (transcripts.fasta)
Number of sense transcripts: 1246 (TGSout.sn)
Number of antisense transcripts: 2086 (TGSout.as)
Number of remaining transcripts: 2667 (TGSout.rem)
Number of sense genes: 2491 (TGSout.sn, TGSout.ffn, TGSout.faa) [NOTE: This is reported as 2159 in the paper. This discrepancy is a result of some bug fixes.]
Number of elements reported in GFF format (sense genes + antisense transcripts): 2491 + 2086 = 4577 (TGSout.gff)

TGS output format
=================
This format lists the coordinates of putative genes. This file consists of five columns (start position, end position, strand, frame, score).  For example,

>ftranscript:1741:5049
217     1059    +       1       1.297925
1061    1993    +       2       1.310458
1994    3280    +       2       1.289984
>ftranscript:6551:6792
1       242     -       3       1.304437


Reference
=========
1. Ismail, W.M., Ye, Y., Tang, H.: Gene finding in metatranscriptomic sequences. BMC Bioinformatics 15(Suppl 9):S8 (2014)
2. Rho, M., Tang, H., Ye, Y.: Fraggenescan: predicting genes in short and error-prone reads. Nucleic acids research 38(20), 191-191 (2010)

License
============
Copyright (C) 2013 Wazim Mohammed Ismail, Yuzhen Ye and Haixu Tang.
You may redistribute this software under the terms of the GNU General Public License.