chicSignificantInteractions

Per viewpoint the significant interactions are detected based on the background model. For each viewpoint file, an output file is created with all recorded significant interactions and a target file. The target file is especially useful in the batch mode context; for two consecutive listed control and treatment viewpoints it merges the significant interactions which can then be used to test for a differential interaction scheme.

chicSignificantInteractions supports two modes to detect significant interactions, either by an x-fold over the average background or a loose p-value. In both cases neighboring significant peaks are merged together and an additional p-value is computed based on the sum of interactions for this neighborhood. Only interactions with a higher p-value (as specified by the threshold –pValue) are accepted as a significant interaction.

An example usage is for single mode is:

$ chicSignificantInteractions –interactionFile interactionFilesFolder/Sox17_FL-E13-5_chr1_1000_2000.bed –referencePoints referencePointsFile.bed –range 20000 40000 –backgroundModelFile background_model.bed –loosePValue 0.5 –pValue 0.01

An example usage is for batch mode is:

$ chicViewpointBackgroundModel –matrices matrix1.cool matrix2.cool matrix3.cool –referencePoints referencePointsFile.bed –range 20000 40000 –outFileName background_model.bed

The main difference between single mode and batch mode is that in single mode the parameter –interactionFile is interpreted as a list of viewpoint files created with chicViewpoint, whereas in batch mode only one file is allowed, containing the file names of viewpoint files (one per line). This file is created by chicViewpoint and the parameter –writeFileNamesToFile. In batch mode, please remember to specify the folder (via –interactionFileFolder) where chicViewpoint wrote the files.

usage: chicSignificantInteractions --interactionFile INTERACTIONFILE
                                   [INTERACTIONFILE ...] --pValue PVALUE
                                   (--xFoldBackground XFOLDBACKGROUND | --loosePValue LOOSEPVALUE)
                                   --backgroundModelFile BACKGROUNDMODELFILE
                                   --range RANGE RANGE
                                   [--outFileNameSuffix OUTFILENAMESUFFIX]
                                   [--interactionFileFolder INTERACTIONFILEFOLDER]
                                   [--targetFolder TARGETFOLDER]
                                   [--outputFolder OUTPUTFOLDER]
                                   [--writeFileNamesToFile WRITEFILENAMESTOFILE]
                                   [--targetFileList TARGETFILELIST]
                                   [--batchMode] [--threads THREADS]
                                   [--fixateRange FIXATERANGE]
                                   [--xFoldMaxValueNB XFOLDMAXVALUENB]
                                   [--peakInteractionsThreshold PEAKINTERACTIONSTHRESHOLD]
                                   [--resolution RESOLUTION]
                                   [--computeSampleNumber COMPUTESAMPLENUMBER]
                                   [--help] [--version]

Named Arguments

--xFoldBackground, -xf

Filter x-fold over background. Used to merge neighboring bins with a broader peak but less significant interactions to a single peak with high significance. Used only for pValue option.

--loosePValue, -lp

loose p-value threshold to filter target regions in a first round. Used to merge neighboring bins with a broader peak but less significant interactions to a single peak with high significance. Used only for pValue option.

Required arguments

--interactionFile, -if

path to the interaction files which should be used for aggregation of the statistics.

--pValue, -p

p-value threshold to filter target regions for inclusion in differential analysis.

--backgroundModelFile, -bmf

path to the background file which is necessary to compute the rbz-score

--range

Defines the region upstream and downstream of a reference point which should be included. Format is –region upstream downstream, e.g. –region 500000 500000 plots 500kb up- and 500kb downstream. This value should not exceed the range used in the other chic-tools.

Optional arguments

--outFileNameSuffix, -suffix

File name suffix to save the results; prefix is the input file name.

Default: “_significant_interactions.bed”

--interactionFileFolder, -iff

Folder where the interaction files are stored. Applies only for batch mode.

Default: “.”

--targetFolder, -tf

Folder where the target files are stored.

Default: “targetFolder”

--outputFolder, -o

Output folder of the significant interaction files.

Default: “significantFiles”

--writeFileNamesToFile, -w

Default: “significantFilesBatch.txt”

--targetFileList, -tl

The file to store the target file names.

Default: “targetList.txt”

--batchMode, -bm

Turn on batch mode. The given file for –interactionFile and or –targetFile contain a list of the to be processed files.

Default: False

--threads, -t

Number of threads (uses the python multiprocessing module).

Default: 4

--fixateRange, -fs

Fixate range of backgroundmodel starting at distance x. E.g. all values greater than 500kb are set to the value of the 500kb bin.

Default: 500000

--xFoldMaxValueNB, -xfnb

x-fold factor to increase the number of precomputed p-values per relative genomic distance. If set to 1, the maximal distance is used.

Default: 10

--peakInteractionsThreshold, -pit

The minimum number of interactions a detected peak needs to have to be considered.

Default: 5

--resolution, -r

Resolution of the bin in genomic units. Values are set as number of bases, e.g. 1000 for a 1kb, 5000 for a 5kb or 10000 for a 10kb resolution.This value is used to merge neighboring bins.

Default: 1000

--computeSampleNumber, -csn

Number of samples to compute together. Applies only in batch mode.

Default: 2

--version

show program’s version number and exit