tcode identifies protein-coding regions in one or more DNA sequences using the fickett TESTCODE statistic. This is based on simple and universal differences between protein-coding and noncoding DNA. The TESTCODE statistic is calculated for windows of a specified size over each input sequence. The results can be output as a standard EMBOSS report file or displayed graphically.
The output reports each window as "Coding", "Noncoding" or "No opinion". Entries marked "No opinion" have a TESTCODE value that falls between the maximum and minimum values required to report a region as noncoding or coding. For the graphical plot, all points above a green horizontal line are determined to be coding regions. Those below a red line are determined to be noncoding. Points between the red and green lines are "no opinion" ones.
This application can assist in determining the probability of a region of nucleic sequence encoding a functional protein.
A window of at least 200 bases is moved over the sequence in steps of 3 bases
Let:
A1 = Number of A's in positions 1,4,7 ... A2 = Number of A's in positions 2,5,8 ... A3 = Number of A's in positions 3,6,9 ...
A position value is determined that reflects the degree to which each base is favoured in one codon position over another, i.e.
Apos = MAX(A1,A2,A3) / MIN(A1,A2,A3)+1
This is done for all 4 bases. The percentage composition of each base is also determined. Eight values are therefore determined, four position values and four composition values. These are then converted to probabilities (p) of coding using a look-up table provided as the data file for the program. The values in this look-up table have been determined experimentally using known coding and noncoding sequences.
Each of the probabilities is multiplied by a weight (w) value (again from the look-up table) for the respective base. The weight value reflects the percentage of the time that each parameter alone successfully predicted coding or noncoding function for the sequences of known function.
The TESTCODE statistic is then:
p1w1 + p2w2 + p3w3 + p4w4 + p5w5 + p6w6 + p7w7 + p8w8
A result of less than 0.74 is probably a non-coding region.
A result equal or greater than 0.95 is probably a coding region.
Anything in between these two values is uncertain.
|
|
The program will ignore ambiguity codes in the nucleic acid sequence and just accept the four common bases. This is a function of the algorithm, and the data tables.
tcode outputs a report format file. The default format is 'table'.
The resulting report file will be given a name relating to the analysed sequence together with the .tcode suffix by default. Should there be no sequence description, the default reverts to outfile.tcode.
tcode optionally outputs a graph to the specified graphics device.
The graphical display is output with the default file name tcode.1. and then the name of the selected graphical display (e.g. png; ps).
The graph indicates the threshold for probably being coding with a green horizontal line and the threshold for probably not being coding with a red horizontal line.
The default data file (look-up table) is Etcode.dat which contains the data from the original paper (1)
# Fickett TESTCODE data # Nuc. Acids Res. 10(17) 5303-5318 # # Position parameter values (last value must be 0.0) 1.9 1.8 1.7 1.6 1.5 1.4 1.3 1.2 1.1 0.0 # # # Content parameter values (last value must be 0.0) 0.33 0.31 0.29 0.27 0.25 0.23 0.21 0.17 0.00 # # # Position probabilities for A,C,G,T respectively 0.94 0.80 0.90 0.97 0.68 0.70 0.88 0.97 0.84 0.70 0.74 0.91 0.93 0.81 0.64 0.68 0.58 0.66 0.53 0.69 0.68 0.48 0.48 0.44 0.45 0.51 0.27 0.54 0.34 0.33 0.16 0.20 0.20 0.30 0.08 0.09 0.22 0.23 0.08 0.09 # # # Content probabilities for A,C,G,T respectively 0.28 0.82 0.40 0.28 0.49 0.64 0.54 0.24 0.44 0.51 0.47 0.39 0.55 0.64 0.64 0.40 0.62 0.59 0.64 0.55 0.49 0.59 0.73 0.75 0.67 0.43 0.41 0.56 0.65 0.44 0.41 0.69 0.81 0.39 0.33 0.51 0.21 0.31 0.29 0.58 # # # Weights for position 0.26 0.18 0.31 0.33 # # # Weights for content 0.11 0.12 0.15 0.14
This file is retrievable using EMBOSSDATA.
Window size is set by default to 200. The algorithm requires sufficient sequence to perform the statistic on. The original paper suggests a minimum window size of 200.
Window stepping increment is set by default to 3. This will ensure the resulting information remains in frame.
The TESTCODE statistic reflects the fact that codons are used with unequal frequency and that oligonucleotides and nucleotides tend to be repeated with a periodicity of three. The original paper reports that the test had been thoroughly proven on 400,000 bases of sequence data: it misclassifies 5% of the regions tested and gives an answer of "No Opinion" one fifth of the time.
In the GCG package, the current (version 10.3) TESTCODE application's apparent interpretation of the algorithm is: MAX(A1,A2,A3) / MIN(A1,A2,A3) The EMBOSS tcode program uses the correct Fickett algorithm equation: MAX(A1,A2,A3) / MIN(A1,A2,A3) + 1 thus any plot using the GCG TESTCODE aplication will be slightly higher than the tcode equivalent.
The program will ignore ambiguity codes in the nucleic acid sequence and just accept the four common bases. This is a function of the algorithm, and the data tables.
SPIN - "Uneven positional base preferences" Staden software. Free to academics, versions for both X and Windows platforms.