Figure 1.
Splicing Array Probe and RT-PCR Primer Design
(A) Probe design of the splicing array. There are three oligonucleotide probes for each intron-containing gene: intron (red), splice-junction (blue), and exon (green). In addition, there are also about 800 probes for intronless genes (yellow). This figure is modified from Clark et al. [15].
(B) Primer design of RT-PCR. Primers are chosen to flank the intron–exon2 junction and the second exon or spliced mRNA.
Figure 2.
Graphical Representation of Designs
(A) In this representation, vertices correspond to target mRNA samples and edges to hybridizations between two samples. By convention, we place the green-labeled sample at the tail and the red-labeled sample at the head of the arrow.
(B) Nested design of the experiment. The effect A is nested in S, and S is in turn nested in V. Note that there are two samples (S) for each mutant, but only one sample for the wild-type.
Figure 3.
Scatterplots of the Logarithm Intensities of Splicing-Related Probes
Points are color-coded as indicated.
Figure 4.
Boxplots of Normalized Ratios of Splicing-Related Probes Stratified by Mutants
Splice-junction and exon probe ratios show a shift from the horizontal zero line in the negative direction, whereas intron probe ratios are centered at zero.
Figure 5.
Scatterplots of Normalized Ratios of Splicing-Related Probes
Points are color-coded by their mutant identity. Gray horizontal and vertical reference lines indicate zero expression ratios.
Table 1.
Summary of the Five Competing DE Models
Table 2.
Degrees of Freedom for the ANOVA Mixed Models
Figure 6.
Analysis of DE Gene Ranking and Selection by DEDS
(A) A comparison of ranking of DE genes by DEDS and individual measures. Plotted are a scatterplot matrix of DE gene rankings by the five models and DEDS using spt5–242 IA indices. Ranks are logged so that correlations of DE genes (low ranks) are more clearly displayed.
(B) Sensitivity of DEDS declarations of DE to choice of scale and distance metric. Genes, ordered according to their DE significance by the raw/Euclidean combination (so that the black points are monotone by definition) are plotted against DEDS q-values for all four scale/distance combinations. The dashed gray line marks the 0.05 q-value threshold.
Table 3.
Number of DE in SJ and IA Indices
Table 4.
QPCR Validation DE Microarray Data
Table 5.
Distribution of DE Genes
Table 6.
Properties of DE Genes with a Positive Fold Change (Average)
Table 7.
Properties of DE Genes with a Negative Fold Change (Average)
Figure 7.
Venn Diagram of DE Genes from Different Mutants
(A) compares DE genes among the three spt5 mutants (spt5–194, spt5–4, and spt5–242). Statistical test shows that the common 43 genes are highly significant, with a p-value < 0.001. In (B), spt5 refers to the 43 common genes among all spt5 mutants. The overlaps between spt5 and ceg1–250 (40, p < 0.001), spt5, and spt4 (8, p< 0.001), spt4, spt5, and ceg1–250 (7, p< 0.001) are all significant.
Table 8.
Yeast Strains
Table 9.
Derivation of Variance Components for Model I