Circular reasoning rather than cyclic expression

Abstract

A response to Combined analysis reveals a core set of cycling genes by Y Lu, S Mahony, PV Benos, R Rosenfeld, I Simon, LL Breeden and Z Bar-Joseph. Genome Biol 2007, 8:R146.

Figure 1

Expression profiles of six yeast genes across multiple cell-cycle microarray time courses. Expression profiles for (a) budding yeast CDC5 and DBF2, and (b) their fission yeast orthologs plo1 and sid2. These four genes are all periodically expressed according to our analysis [13,14] but not according to that of Lu and co-workers [9]. (c) Conversely, fission yeast mcm3 and mcm5 are both periodically expressed according to the analysis of Lu et al. [9] but not according to us [13,14,17]. The information in the panels refers to the experiments from which the data come and the method of cell-cycle arrest; for example 'Cho et al. [1] CDC28' indicates a time-course experiment in which the cells were arrested using a CDC28 mutant. The values on the y-axis on each profile indicate the log₂ ratio between the expression at a given time point compared with the average expression across the profile. The rank scores show that plo1 and sid2 are both among the top 100 cycling genes according to our analysis, whereas mcm3 and mcm5 are among the 3,000 least cycling genes. All plots were obtained from the Cyclebase.org database where further details on the normalization procedure and the scoring scheme can also be found [17,38].

Figure 2

Expression values for MCM6 in humans, budding yeast, and fission yeast. Values are log ratios between synchronized and unsynchronized cells. (a, b) Expression profiles of budding yeast MCM6 under different cell-cycle arrest methods [2,3]. (c, d) Expression of fission yeast mcm6 under different arrest methods [4,5]. (e) Expression of MCM6 in human HeLa cells [7]. Cell-cycle stages are shown underneath each panel. Jensen et al. [14] claim that although human MCM6 is cycling at the transcriptional level, its homologs in budding yeast and fission yeast do not cycle. As (a-d) show, the expression of yeast MCM6 seems more cyclic than that of human MCM6, highlighting the limitations of species-by-species thresholding.

Figure 3

Score distributions for fission-yeast genes that are ranked below the cycling score threshold. The red curve is the distribution of 350 fission-yeast orthologs of cycling budding-yeast genes. The black curve is the distribution of all the other 3,641 fission-yeast genes. Density is the distribution density for each of the different scores. As can be seen, the red curve is highly skewed to the right (higher score). In fact, the difference between the two curves is significant, with a p-value of 0.01 (Wilcoxon rank-sum test). Thus, while orthologs of cycling budding-yeast genes may fall just below the fission-yeast threshold, they are still at least weakly cycling, much more so than expected by chance, indicating that expression is conserved at a much stronger rate than the rate determined by thresholding-based methods.

Figure 4

Comparison of expression score ranks and posterior ranks. (a) The expression score rank and posterior rank for fission-yeast genes. The x-axis is the expression score rank (the lower the rank the more cyclic the gene is determined to be by the scoring method) and the y-axis is the rank based on our method (again, the lower the better). As can be seen, the ranks for most of the genes do not change much. The red dashed line represents the posterior threshold used to select cycling genes, and the green dashed line is the corresponding threshold if only expression scores are used. Almost all genes that are elevated by our method to a cyclic status have a rather high cyclic expression score (though some are not as high as the cutoff for score alone, which is where the two methods differ). Five selected genes are highlighted by red circles. These genes would have been missed if only expression scores were used to determined cyclicity, because their scores would be just below the cutoff. While Jensen et al. [14] do not assign cyclic status to these genes, sam1 was also identified as cycling by Peng et al. [5], SPBC17D11.08 was included in the list by Rustici et al. [4], and rpb9 was identified by both Oliva et al. [6] and Peng et al. [5]. The other two genes, SPBP8B7.26 and rmi1, are missing from all three studies, even though their profiles appear cyclic (not shown). (b-d) Similar plots for (b) budding yeast [2,3], (c) human [7], and (d) Arabidopsis [39].