Differential analysis

MA plots are used for differential analysis of condition 1 (cond1) vs. condition 2 (cond2).
Data (bedGraphs) from two conditions are quantile-normalized first, then put into a MA-plot.
A sliding window of 15% of A axis span is used to calculate z-scores (of M).

Positive z-score indicates over-representation of counts in condition 1.
Negative z-score indicates over-representation of counts in condition 2.

Because bedGraph format uses the sign to indicate strand, results are split into two bedGraphs. Magnitude equals to the absolute value of z-score and sign indicates strand:

• Over-represented in condition 1 (i.e., positive z-scores)
• Over-represented in condition 2 (i.e., negative z-scores)

A third bedGraph is also produced showing only windows centered at peaks of high-density smoothed absolute z-values.

MA plots

M=log(count_cond1/count_cond2), A=log(count_cond1)+log(count_cond2).

Top 30 regions with highest density of difference (z-scores) in counts

Click on region to automatically upload the two bedGraphs with z-scores of differential analysis and start exploring in UCSC browser.

Other measures of (dis)similarity

x-links in condition 1: 208288
x-links in condition 2: 138370
All x-link positions (union): 346314
Positions shared by both conditions (intersection): 344
Jaccard index (intersection/union): 0.0010

x-link position similarity

For each x-link in condition 1 (condition 2), look for nearest x-link in condition 2 (condition 1). Draw histogram of distances to closest neighbors.

• Condition 1: both sides 2201.2, downstream -64268.1, upstream 66469.3
• Condition 2: both sides 2572.3, downstream -11158.8, upstream 13731.1

• Condition 1: both sides 2.5, downstream -32.3, upstream 34.8
• Condition 2: both sides -0.7, downstream -36.6, upstream 35.9

Count similarity

For each x-link position with specific value stored in input bedGraph for condition 1 (condition 2) look if same position in bedGraph for condition 2 (condition 1) has a non-zero value.

k-mers similarity

For k-mer analyses performed with same parameters on both conditions, report on similarity of results.
Similarity is measured by correlation of z-scores of all k-mers, and number of top k-mers shared by both conditions.

RNAmaps similarity

For RNAmap analyses performed with same parameters on both conditions, report on similarity of results.
RNAmaps are compared by normalizing them first to sum 1.0 and then taking the average squared absolute difference (units are squared abs. difference per 1000 positions).