Expansion of the circadian transcriptome in Brassica rapa and genome-wide diversification of paralog expression patterns

Coexpression modules were generated for genes in circadian conditions using two different entrainments (LD and HC). Each module was analyzed for categorical enrichment using Biological Process Gene Ontology annotations. Each tab in this file represents a different LD or HC module. The last tab, titled ‘Differential_Pattern’ represents categorical enrichment results on the list of 1713 genes that show significant pattern changes between LD and HC entrainment conditions.

Gene expression was evaluated under circadian conditions (constant light and temperature) for 48 hr after plants were entrained in one of the two conditions; LD (12 hr light/12 hr dark) or HC (12 hr 20 °C/ 12 hr 12°C). Using the set of cycling genes that were identified in both datasets, DiPALM was used to detect differential patterning and differential median expression between LD and HC conditions. This dataset reports these results. Column 1 are gene accessions, Column 2 are adjusted p-values for differential patterning and Column 3 are adjusted p-values for differential median expression.

WGCNA modules generated with cycling genes from a combined LD and HC dataset were analyzed for depletion or enrichment of multi-copy genes. Significant enrichment of multi-copy genes was observed in five morning-phased modules while depletion was observed in four evening-phased modules. This dataset shows the results of Biological Process Gene Ontology categorical enrichment for the set of multi-copy genes in these five enriched modules (‘MorningCopied’ tab) and the set of multi-copy genes in these four depleted modules (‘EveningCopied’ tab).

The LD and HC circadian time-course datasets were combined into one LDHC time course. B. rapa paralogous pairs were identified and DiPALM was used to detect differential patterning and differential median expression between paralogs. This dataset reports these results. Column 1 are gene accessions, Column2 are adjusted p-values for differential patterning and Column 3 are adjusted p-values for differential median expression.

Predicted target groups from GRNs were compared between Arabidopsis (At) and B. rapa (Br) networks. Sets of 3 genes containing one At Transcription factor (TF) and its two Br orthologs (denoted B. rapa 1 and B. rapa 2) were defined. 256 of these TF sets were found between the At and Br GRNs. Tab ‘GRN_Comparison_Orthologs’ reports results after testing for orthology between predicted target groups of corresponding At and Br TFs. Column 1 (Gene Accessions) denotes the three accessions for TFs in each set. For simplicity, B. rapa 1 was always made to be the Br TF with the most significant overlap in target genes with the At TF. Columns 2, 3, and 4 are the total number of genes in the predicted target groups for At, Br1, and Br2, respectively. Columns 5 and 6 are the number of overlapping genes with the At targets (based on orthology) for Br1 and Br2, respectively. Columns 7 and 8 are the p-values for the significance of overlap with At targets for Br1 and Br2, respectively. Column 9 is the p-value result of a permutation test performed to determine if Br1 is significantly more At-like than Br2. This p-value is meaningless unless Column 7 is also significant. Tab: GRN_Comparison_CNSs reports results after testing for conserved non-coding sequences (CNSs) near genes in predicted target groups of corresponding At and Br TFs. Column 1 (Gene Accessions) denotes the three accessions for TFs in each set. For simplicity, B. rapa 1 was always made to be the Br TF with the most significant overlap in target gene CNSs with the At TF. Columns 2, 3, and 4 are the total number of CNSs associated with genes in the predicted target groups for At, Br1, and Br2, respectively. Columns 5 and 6 are the number of overlapping CNSs with the At targets for Br1 and Br2, respectively. Columns 7 and 8 are the p-values for the significance of CNS overlap with At for Br1 and Br2, respectively. Column 9 is the p-value result of a permutation test performed to determine if Br1 is significantly more At-like than Br2. This p-value is meaningless unless Column 7 is also significant. Tab: Significant_Groups summarizes all TF groups where Br1 is significantly more At-like than Br2 based on either the ortholog or CNS comparison. Column 1 lists the three accession numbers for TFs in each group and is ordered based on the target genes orthology comparison (i.e. Br1 is always the most At-like based on orthology of the predicted TF-target genes with predicted At target genes). Column 2 lists the three accessions for TFs in each group and is ordered based on the target genes CNS comparison (i.e. Br1 is always the most At-like based on CNS overlap of the predicted TF-target genes with predicted At target genes). Column 3 lists the p-values for the significance of overlap with At targets and Br1 targets based on orthology. Column 4 indicates if Br1 is significantly more At-like than Br2 based on target gene orthology. Column 5 lists the p-values for the significance of overlap with At targets and Br1 targets based on CNSs. Column 6 indicates if Br1 is significantly more At-like than Br2 based on target gene CNS overlap. Column 7 indicates the 12 TF groups that were used for the motif enrichment analysis (Figure 4—figure suplement 2).

Gene expression data from control samples in a time course performed in diel conditions (14 hr light 21°C/10 hr dark 18°C) were examined (Greenham et al., 2017). B. rapa paralogous pairs were identified and DiPALM was used to detect differential patterning and differential median expression between paralogs. This dataset reports these results. Column 1 are gene accessions, Column 2 are adjusted p-values for differential patterning, and Column 3 are adjusted p-values for differential median expression.

Gene expression data from Greenham et al., 2017 were examined. Mild drought was imposed and compared to well-watered samples. DiPALM was used to identify differential patterning and differential median expression of genes in response to drought. This dataset reports these results. Column 1 are gene accessions, Column 2 are adjusted p-values for differential patterning, and Column 3 are adjusted p-values for differential median expression.

DiPALM was evaluated using simulated benchmark datasets (Spies et al., 2019). Eight different simulated datasets were generated (rows). Dataset names denote the number of simulated counts, number of simulated replicates, and the number of simulated timepoints. This table also includes the top three performing methods mentioned in Spies et al.: splineTC (Michna et al., 2016), maSigPro (Nueda et al., 2014), and impulseDE2 (Fischer et al., 2018). The method of running a traditional differential expression analysis on each timepoint separately and then taking the most significant timepoint for each gene is also shown (pairwise). Three different DiPALM evaluations were carried out using the pattern-detecting version (kME), the abundance-detecting version (Median) and a combined score using the sum of both.

Using a list of canonical CNS sequences derived from Haudry et al., 2013, the TAIR10 genome was annotated for those CNSs using NCBI BLAST+. This dataset is in BED file format with six fields as follows: Field 1 denotes the chromosome, Field 2 denotes the start nucleotide position, Field 3 denotes the end nucleotide position, Field 4 denotes the CNS accession (name), Field 5 is a placeholder ‘1’ for a score annotation but is not used in this case, Field 6 denotes the strand.

Using a list of canonical CNS sequences derived from Haudry et al., 2013, the B. rapa R500 genome was annotated for those CNSs using NCBI BLAST+. This dataset is in BED file format with six fields as follows: Field 1 denotes the chromosome, Field 2 denotes the start nucleotide position, Field 3 denotes the end nucleotide position, Field 4 denotes the CNS accession (name), Field 5 is a placeholder ‘1’ for a score annotation but is not used in this case, Field 6 denotes the strand.