(3) Dregion
Dregion, the third peak-calling algorithm in DANPOS, analyzes change in width, summit height, and total signal in each enriched region between samples.
Example 1. Define enriched region for histone modification H3K4me3 in a single sample, with input effect subtracted:
Test data (right click to download and put files in directories as indicated below):
(1) H3K4me3_sampleA/
(2) input_sampleA/
Command:
python danpos.py dregion H3K4me3_sampleA -b input_sampleA
Example 2. Compare H3K4me3 ChIP-Seq data between two samples A and B, each sample has its own input effect to be subtracted.
Test data (right click to download and put files in directories as indicated below):
(1) H3K4me3_sampleA/
(2) input_sampleA/
(4) input_sampleB/
Command:
python danpos.py dregion H3K4me3_sampleA:H3K4me3_sampleB.bed.gz -b H3K4me3_sampleA:input_sampleA,H3K4me3_sampleB.bed.gz:input_sampleB
Example 3. Compare H3K4me3 ChIP-Seq data between two samples A and B, with spike-in controls to specify libarary size for each sample, each sample has its own input effect to be subtracted.
Test data (right click to download and put files in directories as indicated below):
(1) H3K4me3_sampleA/
(2) input_sampleA/
(4) input_sampleB/
Spike-ins information:
Suppose that we have spike-ins to indicate that there would be 10 and 20 million reads for sample A and B, respectively.
Command:
python danpos.py dregion H3K4me3_sampleA:H3K4me3_sampleB.bed.gz -b H3K4me3_sampleA:input_sampleA,H3K4me3_sampleB.bed.gz:input_sampleB -c H3K4me3_sampleA:10000000,input_sampleA:10000000,H3K4me3_sampleB.bed.gz:20000000, input_sampleB:20000000
Example 4. Quantile Normalization: Different from DANPOS1, currently, the -n Q parameter for quantile normalization is disabled. If you try to use quantile normalization, first, use dpos function to generate wiq file with -n N parameter, which disable all normalization method and --smooth_width 0 which prevent smooth. Then, you could use "python danpos.py wiq --buffer_size 50 chromosome_size_file yourfileA.wig:yourfileB.wiq" to normalize your file with quantile normalization and call danpos peak caller functions (dpos, dpeak, dregion, dtriple) again. Make sure you always disable -n N and only smooth your data once.