******************************************************************************
Table S5a.txt --- preferred distances of motif pairs.

This file shows the preferred distances between motif pairs in every motif module.

For every motif module, we calculated the distances between motif pairs. For a motif pair, we divided the distances of the instances into k bins, with each bin spanning 50bp. Note that the top 1% distances were put to the same bin. Using a Poisson model, we calculated the p values of significant enrichment in one bin for every motif pair. All p values were corrected using bonferroni correction and motif modules with p values less than 0.01 were selected to have distance preference. 

We obtained 9086692 motif pairs (in 2857010 motif modules) with preferred distance bins. The mean of the distances between motif pairs is 29.83146bp. 

In the supplement tables, each row represents a motif pair. The 1-st column is the name of motif module; the 2-nd column is the motif pair, the next two columns are the preferred distance bin; and the 5-th column is the p value after bonferroni modification. Such as: 1, M00738_M00516,       0, 49, 0, means in the 1-st motif module, the preferred distance bin of M00738 and M00516 is [0,49), and the p value after bonferroni modification is 0.

***************************************************************************
Table S5b.txt --- the preferred distance of CRMs relative to TSSs

This file shows the distance preference of the CRMs relative to TSSs.

Here for every motif module, we calculated the distances between CRMs and the target genes' TSSs. The distances were then divided to several bins, each with length 100bp; note that the top 1% distances were put to the same bin. Using a Poisson model, we calculated the p values of significant preference of a distance for every motif module. All p values were corrected using bonferroni modification and motif modules with corrected p values less than 0.01 were selected to have distance preference. 

We obtained 1217409 motif modules with preferred distance bins. The mean of the distances between CRMs and TSSs is 590.818bp. 

In the supplement tables, each row represents a motif module. The 1-st column is the name of the motif module; the next two columns are the preferred distance bin; and the 4-th column is the p value after bonferroni modification. Such as 4, 136, 235, 3.034343e-06, means in the 4-th motif module, the preferred distance bin is [136,235), and the p value is 3.034343e-06.

*******************************************************************************
Table S5c.txt --- preferred location of motif modules

This file shows the preferred location of motif modules. 

For every motif module, we calculated the number of CRMs falling in genes' upstream, downstream, and introns. Note that CRMs with distance to the TSS farther than 100kb were assigned as "cannot determine". Using a Binomial model, we calculated the p values of location preference for every motif module. All p values were corrected using bonferroni correction and motif modules with p values less than 0.01 were selected to have preferred locations.

We selected 2415634 motif modules. 210744 of them preferred to locate in genes's upstream, 1074481 of them preferred to locate in genes's intron, 1085929 of them preferred to locate in genes's downstream. 

In this file, each row represents a motif pair. The 1-st column is the name of motif module; the 2-nd column is the preferred locations and the 3-rd column is the p value after bonferroni correction.

*******************************************************************************
Table S5d.txt --- strand preference

This file describes the strand preference of the motif modules. 

For every motif module, we examined whether the motifs are in the same strand in all CRMs. For every motif pairs, a p value was assigned using a binomial model. Motif pairs with q value less than 0.01 were selected to have significant strand patterns.

We obtained 111631 motif pairs with significant strand patterns for 107404 motif modules. 41887 of them have opposite strand, and 69774 of them have the same strand. 

In this file, each row represents a motif pair. The 1-st column is the name of motif module, the next two columns are the motif pairs, the 4-th column is the preferred strand (1 means same, -1 means opposite) and the 5-th column is the q value. Such as: 1, M00738, M00516, 1, 0 means in the 1-st motif module, M00738 and M00516 are significantly in the same strand, the q value is 0.