Skip to main content
. 2020 Nov 12;9:e62669. doi: 10.7554/eLife.62669

Figure 5. Association of features with variant causality.

(A) Non-TF features. The figure shows the strength of association between each feature and variant causality. Error bars show the standard error of the mean. Significant associations are indicated by three stars (FDR < 5%) or one star (nominal p-value<0.05). Non-significant features are shown in lighter coloring. (B) TF summary features aggregated across all 196 TFs, separated by strand, mode of aggregation across TFs, strength of binding (weak or strong), and mode of comparing allelic PWM scores across sliding sequence windows spanning each variant (Materials and methods). Each of these summary features was significantly associated with variant causality at an FDR of <5%. (C) Distributions of logistic regression estimates for strong vs. weak binding for individual TFs. The p-value shows the result of a Wilcoxon rank test. See also Figure 5—figure supplement 1.

Figure 5—source data 1. Single-feature regression analyses of causal variants.
Logistic and linear regression (to predict log-fold change) are in separate worksheets.
elife-62669-fig5-data1.xlsx (384.1KB, xlsx)

Figure 5.

Figure 5—figure supplement 1. Examples of predicted TFBS changes at individual variants.

Figure 5—figure supplement 1.

(A) A variant in the promoter of the SFA1 gene alters a strong Msn2/4 motif (Yeastract consensus motif: CCCCT); as well as a strong Haa1 motif (SMGGSG). TFBSs detected by the Yeastract website in the given sequence are shown as colored arrows indicating the strand on which the motif match was detected. TFBSs above the sequence were detected for the BY allele, while those below the sequence were detected for the RM allele. Two TFBSs that differ between the alleles are shown as stronger lines. The histogram shows the distribution of change in TFBS score to weak TFBSs. The dashed yellow line indicates the median change in TFBS score. TFBS scores correspond to log2(predicted likelihood of binding). (B) A variant in the promoter of the HAL5 gene that did not alter strong TFBSs. Nevertheless, this variant caused a similar change in gene expression, accompanied by a similar median change to weak TFBSs, as the variant in (A).