Number of coding ORFs in the yeast genome

1. S. Cebrat, M.R. Dudek, A. Rogowska, 1997, Asymmetry in nucleotide composition of sense and antisense strands as a parameter for discriminating open reading frames as protein coding sequences. Journal of Applied Genetics, 38(1), 1 - 9. (abstract)

2. S. Cebrat, M. R. Dudek, P. Mackiewicz, M. Kowalczuk, M. Fita, 1997, Asymmetry of coding versus non-coding strand in coding sequences of    different genomes. Microbial & Comparative Genomics. 2(4), 259 - 268. (abstract)

3. S. Cebrat, M. Dudek, P. Mackiewicz, 1997, Is there any mystery of ORPHANs? Journal of Applied Genetics,38(4), 365 - 372. (abstract)

4. S. Cebrat, P. Mackiewicz, M. R. Dudek, 1998, The role of the genetic code in generating new coding sequences inside existing genes. Biosystems, 45(2), 165 - 176. (abstract)

5. S. Cebrat, M. R. Dudek, P. Mackiewicz, 1998, Sequence asymmetry as a parameter indicating coding sequence in Saccharomyces cerevisiae genome. Theory in Biosciences, 117, 78 - 89. (abstract)

6. M. Kowalczuk, P. Mackiewicz, A. Gierlik, M. R. Dudek, S. Cebrat, 1999, Total Number of Coding Open Reading Frames in the Yest Genome. YEAST, 15, 1031-1034. (abstract)

7. P. Mackiewicz, M. Kowalczuk, A. Gierlik, M. R. Dudek, S. Cebrat, 1999, Origin and properties of noncoding ORFs in the yeast genome.Nucleic Acids Res. 27(17), 3503-3509.(abstract)

8. P. Mackiewicz, M. Kowalczuk, D. Mackiewicz, A. Nowicka, M. Dudkiewicz, A. Laszkiewicz, M. R. Dudek, S. Cebrat,
2002, How many protein-coding genes are there in the Saccharomyces cerevisiae genome? Yeast 19(7), 619-629.



    Protein coding ORFs have specific nucleotide composition, which allows to distinguish between coding and noncoding sequences (Cebrat et al., 1997a, 1997b, 1998). Compositionl trends in an ORF can be depicted by a DNA walk in two dimensional space (see  Graphic representation of coding DNA sequences - a spider). To measure compositional bias of protein coding sequences, we have calculated arcus tangent of (G-C)/(A-T) for the first and the second codon positions of yeast ORFs. This parameter measures the departure from  A=T and G=C equimolarities within the sense strand and allows us to present individual Open Reading Frames as single points on the finite surface of a torus (see Distribution of ORFs in a torus projection). Distribution of genes with described phenotypes is much more compact than distribution of all ORFs. That means that all genes that have been described so far have different nucleotide composition than ORFs with unknown functions (Cebrat et al., 1998).

    By comparing distribution of all ORFs to distribution of the set of genes with known function, we have estimated the total number of protein coding intronless ORFs longer than 100 codons in the yeast genome at about 4700 (Cebrat et al., 1997a, 1998). That estimation was confirmed for 4800 two years later using a larger pool of known genes (Kowalczuk et al., 1999). This number should not be directly compared with the estimations of the number of all coding sequences done by other authors (which was done many times, leading to misunderstanding) because our number has referred only to intronless ORFs longer than 100 codons and not to all ORFs annotated in database (including intron-containing ORFs and ORFs shorter than 100 codons).
    In the paper (Mackiewicz et al., 2002) we have approximated by two independent methods the total number of protein coding sequences among all sequences annotated in MIPS (including genes with introns and ORFs shorter than 100 codons as well) for 5300-5400. Moreover, in this paper we have compared the results of estimations of the total number of protein coding genes in the Saccharomyces cerevisiae genome, which have been obtained by many laboratories since the yeast genome sequence was published (MIPS database, Zhang and Wang, 2000, Souciet et al., 2000, Blandin et al., 2000, Malpertuy et al., 2000; Wood et al., 2001). It seems that the total number of protein coding ORFs in this genome is several hundred lower than originally assumed (Goffeau et al., 1996; Winzeler & Davis, 1997; Mewes et al., 1997).
    However, it explains the "mystery of orphans", which is the large number of ORFs with unknown function and no homology to any known genes, discovered during systematic sequencing of the yeast genome (Dujon, 1996). According to us, most orphans do not code for proteins (Cebrat et al., 1997c). Nucleotide composition of many of them resembles the antisense of genes, so they may have been generated by coding sequences in the past and later moved to intergenic space by duplication (Mackiewicz et al., 1999, Gierlik et al., 1999). We have found about 700 ORFs in the MIPS database whose putative protein products have homologues in frames different than the frame which had been assumed in the database as coding (Mackiewicz et al., 1999).

Combined data of coding probabilities and compositional parameters counted by us for each ORFs and some data published by other authors are available below.

Tabelarized data:
 
Chromosome I
Chromosome II
Chromosome III
Chromosome IV
Chromosome V
Chromosome VI
Chromosome VII
Chromosome VIII
Chromosome IX
Chromosome X
Chromosome XI
Chromosome XII
Chromosome XIII
Chromosome XIV
Chromosome XV
Chromosome XVI

Tha data for all chromosomes in Excel format: chr1_16.xls
                                                                  chr1_16.zip

Tables description:
 
chr. 
ORF's name
start
stop
length
S1
S2
V1
V2
D
cod. prob.
YZ score
MIPS
Genol.
T
Wood's annotation
brief ID

 
chr.chromosome number;
length length in codons;
parameters of DNA asymmetry:
S1, S2values of angles (in degrees) and V1, V2normalized length of vectors, for the first and the second codon positions respectively;
Dthe Euclidean distance of the ORF from the center of known genes’ distribution in the four-dimensional space of parameters S1, S2, V1, V2;
cod. prob. - coding probability (Mackiewicz et al.., 2002); http://smORFland.uni.wroc.pl;
YZ scorecoding probability according to Zhang and Wang (2000);
MIPSclass in MIPS database (http://www.mips.biochem.mpg.de/proj/yeast/):
1-known protein;
2-strong similarity to known protein (higher than one third of FASTA self-score);
3-similarity to known protein (lower than one third of FASTA self-score);
4-similar to unknown protein;
5-no similarity;
6-questionable ORF;
Similarities have been measured by FASTA scores. A FASTA score between 100 and 200 was defined as "weak similarity". FASTA scores between 200 and 1/3 of the selfscore (FASTA score of the protein, when aligned with itself) of the protein were defined as "similarity". A FASTA score over 1/3 of selfscore was defined as "strong similarity". A questionable ORF is defined by a combination of the following attributes: low CAI value, partial overlap to a longer or known ORF, no similarity to other ORFs.
Genol.class in Genolevures program (Souciet et al., 2000, Malpertuy et al., 2000), http://cbi.labri.u-bordeaux.fr/Genolevures/Genolevures.php3:
0-probably spurious ORF;
1-ORF conserved in non-Ascomycetes;
2-ORF conserved in Ascomycetes only;
3-ORF without homology in orgs. other than S. cerevisiae itself;
T - total number of yeast species in which at least one homologue of the S. cerevisiae ORF has been identified by Genolevures program;
Wood’s annotationre-annotations of ORFs done by Wood et al. (2001); http://www.sanger.ac.uk/Projects/S_cerevisiae.

Parameters describing asymmetry of coding sequences:

To visualise asymmetry of DNA sequence and to show the biological meaning of the parameters used, DNA walks are performed (Cebrat et al., 1997a, 1997b, 1998). Imagine that a virtual walker starts its walk on the ORF sequence from the first nucleotide of the start codon. Next, it jumps to the first nucleotide in the second codon and so on (Fig.1A). It stops at the first nucleotide of the last codon of the analysed ORF. These walks (or jumps) are translated into a plot in a two-dimensional space where the walker goes one unit up if the visited nucleotide is guanine, down if the visited nucleotide is cytosine, right if adenine and left if thymine (Fig. 1B). Then, the walker does its walk for the second and the third codon positions. Since there are very strong and specific compositional trends in each position in coding sequences, the plots “drawn” by the walker are also specific for each position in codons and do not resemble Brownian motion (Fig. 1C).
In this study, we have used two pairs of parameters describing the obtained walks, which in fact are measures of the asymmetry in composition of the first and the second positions in codons of ORF sequences:
           (eq. 1)
       (eq. 2)
where A, T, G, and C are the numbers of respective nucleotides in the first or second positions in codons, and N is the length of the analyzed ORF in codons.)
The parameter S (Fig. 1B) represents the angle between x-axis and the vector determined by the beginning and the end of the corresponding walk (eq. 1). It represents the relation between relative abundance of purines over pyrimidines in complementary pairs of nucleotides. To avoid infinite values of slopes, we have used a measure of angle rather than tangent. Furthermore, the function of arctangent in many instances “normalises” distributions. The parameter V (Fig. 1B) describing the asymmetry was the length of the vector described by the beginning and the end of the walk (eq. 2). The length of vector representing a sequence was divided by square root of the length of this sequence in codons. Such normalisation shows the relation between the length of this vector and the average vector for random DNA sequence of the same length.
The parameter S allows us to present individual Open Reading Frames as single points on the finite surface of a torus (see Distribution of ORFs in a torus projection). In this case angle 1 corresponds to X-coordinate and angle 2 to Y-coordinate. If you want to prepare ORFs distribution of a chromosome, just plot the S1 values against S2 values. Then you will find the position of any particular ORF in the distribution of all ORFs of the chromosome.
 
 
Figure 1. The method of DNA walks on coding sequence:

(A) three DNA walks performed separately for each position in codons beginning from start codon of ORF;
 
 

(B) two-dimensional representation of the above three DNA walks. Two pairs of parameters describing asymmetry of the first and the second positions in codons have been shown: 
S - angle between a vector representing the walk and x-axis, 
V - length of this vector;
 
 
 
 
 
 

(C) two-dimensional representation of three DNA walks performed for the gene SNF12 (YNR023w) coding component of SWI/SNF global transcription activator complex.

If you have any questions, do not hesitate to contact cebrat@smorfland.uni.wroc.pl