20190925 note

[chtsai0105]

• [x] Confirm which kind of statistical analysis was applied in delta data. (MMU or chi-square) -- Confirmed is chi-square test
• [x] Apply color code in MA-plot.
• [ ] Expand delta data from low-res (bind of unbind) to medium-res. (binding amount)
• [ ] Compare with CK's PSI distribution results.
• [x] Promoter region-wise zero out.
• [ ] Check splicing pattern between KLF14 highly expressed and low expressed in out.

[chtsai0105]

MA-plot with color code MA-plot with gray color

[jiahsinhuang]

• In order to find out how Promoter region-wise zero out showed such counterintuitive results, please provide the proportional distribution of the binding differences across 2500bp for each TF.

[jiahsinhuang]

• Expand delta data from low-res (bind or unbind) to medium-res. (binding amount) I suggest to use the simple way to do this: summation of all "1 (difference between two events)" across the promoter and divided by the length of its motif and then round up decimal numbers to zero decimal places.

[chtsai0105]

Promoter wise zero-out showed that the features at 2500-2000 bp upstream TSS have more impact on splicing status changes

[hktsai]

To be honest, I would think the figure is in the reversed direction (X axis means the distance to TTS)

[zjin1126]

This is the TF occupancy along with the promoter position (from -2000 to +500). The TSS site has the highest occupancy, just as most biologists imagination. -2000 to -1500 has a high importance score but a relative low TF occupancy. So far I don't know how to explain this result.