Analysis of codon pair frequencies in genomic sequence
Buchan, JR, Aucott, L. and Stansfield, I
Introduction
Perl version 5.005_3 was used to run all programs described.
The E.coli geneome sequence has been supplied for test purposes in a trimmed format (a sample
of 100 ORFs only in TIGRecoli-trim.txt). The genome has been reformatted to remove all hard
returns (paragraph marks) except those that separate one sequence from another.
Sequences are in the format:
>sequence name1 ###ATGNNNNNNNNNNNNNNNNNNNN hr
>sequence name2 ###ATGNNNNNNNNNNNNNNNNNNNN hr
Where: ‘ATG’ represents the first codon of each open reading frame, and ‘###’ represents marker
text defining the start of the open reading frame . hr denotes a hard return. ‘N’ denotes any
nucleotide.
In the instructions below, bold font indicates a command to enter in a DOS window whilst italicised
comments represent the name of a file (e.g. programs file, spreadsheet, sequence file). Menu
commands are underlined.
Generating codoncount data using E.coli as an example
The program codoncount.txt counts all the codon pairs in a genome ORF by ORF and records the
total observed count of each of the 3904 pair types (61 sense codons x 64 sense and stop
codons). It also calculates the expected number of codon pairs as each ORF is processed, and
records a cumulative expected codon pair count for the whole genome.
Results are stored in the file named ‘results.txt’, and can be opened in Excel as a tabdelimited .txt file. In Excel the spreadsheet will comprise an upper block of data 64 rows X 130
columns (starts in cell A3). This is the observed codon pair data while the lower 64R X 130C block
(starts in cell A68) is the expected codon pair count data.
The data can be converted into a more manageable form using the spreadsheet
‘Codoncount clean.xls’ Select the entire results.txt worksheet, copy it, and paste the data into the
‘Codoncount RAW’ worksheet of Codoncount clean.xls at cell A1. This data is then referenced in
the ‘Cleaned data’ worksheet of Codoncount clean.xls such that observed codon pair data (top)
and expected codon pair data bottom is listed in 65R x 62C blocks. 5’ P site codons are listed in
rows 3 (observed) and 71(expected) and 3’ A site codons are listed in column A.
Analysing codoncount data in other spreadsheets
Data in Codoncount clean.xls can now be directly pasted into the spreadsheet Residual
calcualtor.xls. This normalises expected count data for amino acid pair bias and calculates
residual scores for ever codon pair. Copy the observed codon pair data (cells A1:BJ67) from
Codoncount clean.xls and Paste special values into the Residual calcualtor.xls ‘observed’
worksheet at cell A1. Copy the observed expected pair data (cells A69:BJ165) from Codoncount
clean.xls and Paste special values into Residual calcualtor.xls ‘expected’ worksheet at cell A1.
Codonpairindex
The program codonpairindex.txt, calculates the mean codon pair residual score (i.e. the Codon
Pair index) for each ORF in a genome.
The program is first ‘loaded’ with a library of residual scores for each potential codon pair.
The supplied version of the program contains the E.coli codon pair residual library. The filename
for the genome being analysed must be accurately referenced in codonpairindex.txt (version
supplied refers to TIGRecoli-trim.txt).
Once the program is run, the mean CPI score for each ORF is written to the file results.txt.
This can be opened in Excel.
Random
Random.txt is a program that generates a randomised genome for a given organism based on the
codon usage values of that organism.
Random.txt is set to generate random ORFs with an E.coli codon bias. This is achieved by
selecting a random number between 0 and 1. Each codon is assigned to a discrete band of values
within this range. The width of that band is proportional to that codon’s biased usage among the 61
sense codons of the genetic code. An ‘if’ statement then determines which of the 61 sense
codons will be chosen in response to each random number generation. Gene size and the number
of genes generated are specified by editing the program at lines 2 and 5. The text “###” is
introduced prior to the gene sequence so that the output genome file is ready for analysis by the
programs listed above.