3.2 第二種方式MAnorm差異分析-一款尋找兩個ChIP-Seq樣本之間差異peak的軟件
軟件網址：https://manorm.readthedocs.io/en/latest/usage.html#id7
使用參考：http://www.itdecent.cn/p/a1a17c42946f
通過比較兩個樣品的common peak的density差異，標準化unique peaks，也就是說，既然兩個樣本間common peak強度一致，那么peak內的reads差異倍數(shù)就是測序深度/密度的差異，能夠作為normalization的標準。直接比較標準化后的peaks，避免了不同樣品信噪比不同的問題。這個算法基于這樣的假設：兩個樣本間都有的 peak 或是 banding 位點，相關蛋白的結合機制相同，故應有相同的 binding intensity。
3.2.1 MAnorm安裝
git clone https://github.com/shao-lab/MAnorm.git #安裝最新版本
cd MAnorm
pip install . #注意.不要漏掉！
manorm --version ##檢查一下是否安裝成功，我的顯示1.3.0
#注意：我是在conda中chipseq環(huán)境中安裝的，所以我要在chipseq環(huán)境中應用
3.2.2 MAnorm使用
manorm --p1 peaks_file1.bed --p2 peaks_file2.bed --pf macs3 --r1 reads_file1.bed --r2 reads_file2.bed
--rf bam --n1 name1 --n2 name2 -o output_dir
#--p1 Peak file of sample 1. (可用MCAS3的結果，如sample1_peaks.xls)
#--p2 Peak file of sample 2
#--pf, --peak-format Format of the peak files. Default: bed,我們用的macs結果，所用這里是macs
#--r1 Read file of sample 1.
#--r2 Read file of sample 2.
#--rf Format of the read files. Default: bed
#--n1 Name of sample 1.
#--n2 Name of sample 2.
#-o output_dir #指定輸出文件路徑
#peak和Read文件格式https://manorm.readthedocs.io/en/latest/usage.html#peak-file-formats
#額外的參數(shù)：--s1, --shiftsize1 Single-end reads shiftsize of sample 1. Default: 100
--s2, --shiftsize2 Single-end reads shiftsize of sample 2. Default: 100
--pe, --paired-end Paired-end mode.
-w, --window-size Window size to count reads and calculate read densities. Default: 2000
--summit-dis Summit-to-summit distance cutoff for common peaks. Default: -w/4
--n-random Number of simulations to test the enrichment of peaks overlap between two samples.
-m, --m-cutoff Absolute M value (log2-ratio) cutoff to define biased (differential binding) peaks.
-p, --p-cutoff P value cutoff to define biased peaks.
--wa, --write-all Output additional files which contains the results of original (unmerged) peaks.
3.2.2 MAnorm結果文件
1. <name1>_vs_<name2>_all_MAvalues.xls #主要結果文件
This is the main output result of MAnorm which contains the M-A values and normalized read densities of each peak, common peaks from two samples are merged together.
chr: chromosome name
start: start position of the peak
end: end position of the peak
summit: summit position of the peak (absolute position)
m_value: M value (log2 fold change) of normalized read densities under comparison
a_value: A value (average signal strength) of normalized read densities under comparison
p_value
peak_group: indicates where the peak is come from and whether it is a common peak
normalized_read_density_in_<name1>
normalized_read_density_in_<name2>
Coordinates in .xls file is under 1-based coordinate-system.\
2.output_filters/
This folder contains the filtered biased/unbiased peaks in BED format.
<name1>_vs_<name2>_M_above_<m_cutoff>_biased_peaks.bed
<name1>_vs_<name2>_M_below_-<m_cutoff>_biased_peaks.bed
<name1>_vs_<name2>_unbiased_peaks.bed
3. output_tracks/
These files are genome track files for M values, A values and P values in wiggle format, you can load these files into a genome browser for visualization.
<name1>_vs_<name2>_M_values.wig
<name1>_vs_<name2>_A_values.wig
<name1>_vs_<name2>_P_values.wig
4.output_figures/
This folder contains M-A plots before/after normalization and a scatter plot which shows the scaling relationship between two samples.
<name1>_vs_<name2>_read_density_on_common_peaks.pdf
<name1>_vs_<name2>_MA_plot_before_normalization.pdf
<name1>_vs_<name2>_MA_plot_after_normalization.pdf
<name1>_vs_<name2>_MA_plot_with_P_value.pdf
