Asymmetry indices for analysis and prediction of replication origins in eukaryotic genomes

Abstract

DNA replication was recently shown to induce the formation of compositional skews in the genomes of the yeasts Saccharomyces cerevisiae and Kluyveromyces lactis. In this work, I have characterized further GC and TA skew variations in the vicinity of S. cerevisiae replication origins and termination sites, and defined asymmetry indices for origin analysis and prediction. The presence of skew jumps at some termination sites in the S. cerevisiae genome was established. The majority of S. cerevisiae replication origins are marked by an oriented consensus sequence called ACS, but no evidence could be found for asymmetric origin firing that would be linked to ACS orientation. Asymmetry indices related to GC and TA skews were defined, and a global asymmetry index I(GC,TA) was described. I(GC,TA) was found to strongly correlate with origin efficiency in S. cerevisiae and to allow the determination of sets of intergenes significantly enriched in origin loci. The generalized use of asymmetry indices for origin prediction in naive genomes implies the determination of the direction of the skews, i.e. the identification of which strand, leading or lagging, is enriched in G and which one is enriched in T. Recent work indicates that in Candida albicans and in several related species, centromeres contain early and efficient replication origins. It has been proposed that the skew jumps observed at these positions would reflect the activity of these origins, thus allowing to determine the direction of the skews in these genomes. However, I show here that the skew jumps at C. albicans centromeres are not related to replication and that replication-associated GC and TA skews in C. albicans have in fact the opposite directions of what was proposed.

Figure 1. GC and TA skew jumps at replication origins (A) and at termination sites (B) in Saccharomyces cerevisiae.

Average values of GC and TA skews computed for third codon positions (thick, red line and thick, blue line, respectively) and for intergenes (thin, red line and thin, blue line, respectively) were calculated every 500 bp over a sliding window of 10 kb. Position 0 corresponds to the position of the extended ACS sequence 5′-(T/A)(T/G)TTTAT(G/A)TTT(T/A)(G/C)(T/G)T-3′ for origins (A) and to the midpoint of termination regions (B). Average GC and TA skew values related to protein coding were substracted from the total GC and TA skew values calculated for third codon positions, so as to plot on the same graph the skew curves corresponding to intergenes and to third codon positions. ACS orientation is marked by an arrow in (A).

Figure 2. Schematics illustrating the computation of asymmetry indices.

(A) Definition of the four strand segments. (B) Four simple cases are shown for different positions x. Segments synthesized as leading and lagging strands are shown as blue and red lines, respectively. The small, black rectangles represent coding sequences with their transcriptional orientation given by the associated arrows. The large, black arrowheads indicate the orientation of the replication forks.

Figure 3. Variations of asymmetry indices around replication origins (A) and termination sites (B) in Saccharomyces cerevisiae.

The average values of I_GC,cod (thick, red line), I_TA,cod (thick, blue line), I_GC,int (thin, red line) and I_TA,int (thin, blue line) were computed every 500 bp using a window length L = 10 kb. Position 0 corresponds to the position of the extended ACS sequence 5′-(T/A)(T/G)TTTAT(G/A)TTT(T/A)(G/C)(T/G)T-3′ for origins (A) and to the midpoint of termination regions (B). All origins were oriented according to their ACS, as indicated by the arrow in (A).

Figure 4. Origin efficiency correlates with the global asymmetry index I_GC,TA in Saccharomyces cerevisiae.

(A,B) Box plots displaying the differences in I_GC,TA values between replication origins annotated as confirmed, likely, and dubious (A) or as chromosomically active and inactive (B). The horizontal lines show the median values. The bottoms and tops of the boxes show the 25th and the 75th percentiles, respectively. (C,D) I_GC,TA values are plotted as a function of origin efficiency and average replication time, respectively. The dashed lines correspond to least square fits.

Figure 5. ROC curves displaying the discriminative power of the asymmetry indices I_GC,cod, I_TA,cod, I_GC,int and I_TA,int.

(A) Intergenes were classified according to their values of I_GC,cod (thick, red line), I_TA,cod (thick, blue line), I_GC,int (thin, red line) and I_TA,int (thin, blue line). (B) Intergenes were classified according to their values of I_GC,cod+I_TA,cod (thin, red line), I_GC,int+I_TA,int (thin, blue line) and I_GC,TA (thick, red line). The ROC curve corresponding to I_GC,cod (thin, black line) is shown for comparison.

Figure 6. Variations of GC skew around Candida albicans centromeres (A) and around Saccharomyces cerevisiae replication origins (B).

(A) Average values of GC skew were computed every 200 bp using sliding windows of 1.5 kb and taking into account all sequences, as described in . (B) Average GC skew values were computed every 200 bp using sliding windows of 1.5 kb and taking into account third codon positions. The horizontal lines correspond to the mean GC skew values, averaged over all positions. The dashed, vertical lines mark positions −3 and 3 (kb) for C. albicans and positions −20 and 20 (kb) for S. cerevisiae.

Figure 7. Variations of average GC and TA skews across interorigin intervals in Candida albicans.

The skews, given in percent, were computed for third codon positions (A) and for intergenes (B). The lines correspond to the fits determined using quasibinomial models (see Materials and Methods).