// Copyright (c) 1997-99 by Arthur Delcher, Steven Salzberg, Simon
// Kasif, and Owen White. All rights reserved. Redistribution
// is not permitted without the express written permission of
// the authors.
// Version: 1.1 April 2003 (S. Salzberg)
// Compute the optimal length for minimum "long"
// orfs, so that the program will return the largest
// number of orfs possible. The -g switch still works
// if specified, but I don't know why anyone would want
// to use that for a training set.
// Also, change min overlap by default to be 0.
// Version 1.04 revised 10 May 99 to add -l command-line switch
// to both glimmer and long-orfs to regard genome as *NOT*
// circular. Default is to regard it as circular.
Program long-orfs takes a sequence file (in FASTA format) and
outputs a list of all long "potential genes" in it that do not
overlap by too much. By "potential gene" I mean the portion of
an orf from the first start codon to the stop codon at the end.
The first few lines of output specify the settings of various
parameters in the program:
Minimum gene length is the length of the smallest fragment
considered to be a gene. The length is measured from the first base
of the start codon to the last base *before* the stop codon.
This value can be specified when running the program with the -g option.
By default, the program now (April 2003) will compute an optimal length
for this parameter, where "optimal" is the value that produces the
greatest number of long ORFs, thereby increasing the amount of data
used for training.
Minimum overlap length is a lower bound on the number of bases overlap
between 2 genes that is considered a problem. Overlaps shorter than
this are ignored.
Minimum overlap percent is another lower bound on the number of bases
overlap that is considered a problem. Overlaps shorter than this
percentage of *both* genes are ignored.
The next portion of the output is a list of potential genes:
Column 1 is an ID number for reference purposes. It is assigned
sequentially starting with 1 to all long potential genes. If
overlapping genes are eliminated, gaps in the numbers will occur.
The ID prefix is specified in the constant ID_PREFIX .
Column 2 is the position of the first base of the first start codon in
the orf. Currently I use atg, and gtg as start codons. This is
easily changed in the function Is_Start () .
Column 3 is the position of the last base *before* the stop codon. Stop
codons are taa, tag, and tga. Note that for orfs in the reverse
reading frames have their start position higher than the end position.
The order in which orfs are listed is in increasing order by
Max {OrfStart, End}, i.e., the highest numbered position in the orf,
except for orfs that "wrap around" the end of the sequence.
When two genes with ID numbers overlap by at least a sufficient
amount (as determined by Min_Olap and Min_Olap_Percent ), they
are eliminated and do not appear in the output.
The final output of the program (sent to the standard error file so
it does not show up when output is redirected to a file) is the
length of the longest orf found.
Specifying Different Start and Stop Codons:
To specify different sets of start and stop codons, modify the file
gene.h . Specifically, the functions:
Is_Forward_Start Is_Reverse_Start Is_Start
Is_Forward_Stop Is_Reverse_Stop Is_Stop
are used to determine what is used for start and stop codons.
Is_Start and Is_Stop do simple string comparisons to specify
which patterns are used. To add a new pattern, just add the comparison
for it. To remove a pattern, comment out or delete the comparison
for it.
The other four functions use a bit comparison to determine start and
stop patterns. They represent a codon as a 12-bit pattern, with 4 bits
for each base, one bit for each possible value of the bases, T, G, C
or A. Thus the bit pattern 0010 0101 1100 represents the base
pattern [C] [A or G] [G or T]. By doing bit operations (& | ~) and
comparisons, more complicated patterns involving ambiguous reads
can be tested efficiently. Simple patterns can be tested as in
the current code.
For example, to insert an additional start codon of CAT requires 3 changes:
1. The line
|| (Codon & 0x218) == Codon
should be inserted into Is_Forward_Start , since 0x218 = 0010 0001 1000
represents CAT.
2. The line
|| (Codon & 0x184) == Codon
should be inserted into Is_Reverse_Start , since 0x184 = 0001 1000 0100
represents ATG, which is the reverse-complement of CAT. Alternately,
the #define constant ATG_MASK could be used.
3. The line
|| strncmp (S, "cat", 3) == 0
should be inserted into Is_Start .
To compile the program:
g++ long-orfs.c -lm -o long-orfs
Uses include files delcher.h gene.h
To run the program:
long-orfs genome.seq
where genome.seq is a genome sequence in FASTA format.
Options can be specified after the genome file name
long-orfs genome.seq <options>
Options are:
-g n Set minimum gene length to n. Default is to compute an
optimal value automatically. Don't change this unless you
know what you're doing.
-l Regard the genome as linear (not circular), i.e., do not allow
genes to "wrap around" the end of the genome.
This option works on both glimmer and long-orfs .
The default behavior is to regard the genome as circular.
-o n Set maximum overlap length to n. Overlaps shorter than this
are permitted. (Default is 0 bp.)
-p n Set maximum overlap percentage to n%. Overlaps shorter than
this percentage of *both* strings are ignored. (Default is 10%.)
If you *DON'T* want to eliminate overlapping genes, just use the -p 100
option.