# Post process Estimation of binding site positions obtained from PING

### Description

Post process Estimation of binding site positions obtained from PING. Refit mixture models with stronger prior in candidate regions contain potential problems, and then convert final result into dataframe.

### Usage

1 |

### Arguments

`ping` |
A 'pingList' object containing estimation of nuclesome positions, result of 'PING' function. |

`seg` |
An object of class 'segmentReadsList' containing the results for all regions pre-processed, 'segmentReads' function. |

`paraEM` |
A list of parameters for the EM algorithm. The default parameters should be good enough for most usages. |

`minK`

:An

`integer`

, default=0. The minimum number of binding events per region. If the value is 0, the minimum number is automatically calculated.`maxK`

:An

`integer`

, default=0. The maximum number of binding events per region. If the value is 0, the maximum number is automatically calculated.`tol`

:A

`numeric`

, default=1e-4. The tolerance for the EM algorithm.`B`

:An

`integer`

, default=100. The maximum number of iterations to be used.`mSelect`

:A character string specifying the information criteria to be used when selecting the number of binding events. Default="AIC3"

`mergePeaks`

:A

`logical`

stating whether overlapping binding events should be picked. Default=TRUE`mapCorrect`

:A

`logical`

stating whether mappability profiles should be incorporated in the estimation, i.e: missing reads estimated. Default=TRUE

`paraPrior` |
A list of parameters for the prior distribution. The default parameters should be good enough for most usages. |

`xi`

:An

`integer`

. The average DNA fragment size.`rho`

:An

`integer`

. A variance parameter for the average DNA fragment size distribution.`alpha`

:An

`integer`

. First hyperparameter of the inverse Gamma distribution for sigma^2 in the PICS model`beta`

:An

`integer`

. Second hyperparameter of the inverse Gamma distribution for sigma^2 in the PING model`lambda`

:An

`integer`

. The lambda control Gaussian Markov Random Field prior on the distance of adjacent nucleosomes, we do not recommend user change the default value.`dMu`

:An

`integer`

. Our best guess for the distance between two neighboring nucleosomes.

`rho2, sigmaB2, alpha2, beta2` |
Integer values, the parameters in the prior of mixture models to be re-fitted. |

`min.dist` |
The minimum distance of two adjacent nucleosomes predicted from different candidate regions, smaller than that will be treated as duplicated predictions for the same nucleosomes. |

`score` |
A |

`dataType` |
A character string that can be set to use selected default parameters for the algorithm. |

`nCores` |
An |

`makePlot` |
A |

`FragmentLength, mart, seg.boundary, DupBound, IP, datname` |
Plotting parameters and options. |

`IP`

:A

`GRanges`

object. The reads used in segmentation process.`FragmentLength`

:An

`integer`

. The length of XSET profile extension

### Value

A `data.frame`

containing the estimation of binding site positions.

### Note

Based on our experiemt on a few real data sets, we suggestion to use following values of parameters. For sonication data we use rho1=1.2; sigmaB2=6400;rho=15;alpha1=10; alpha2=98; beta2=200000. For MNase data we use rho1=3; sigmaB2=4900; rho=8; alpha1=20; alpha2=100; beta2=100000. The value of xi depends on specy of sample, since that affect the length of linker-DNA. For example, we use xi=160 for yeast and xi=200 for mouse.

### Author(s)

Xuekui Zhang <xzhang@stat.ubc.ca>, Sangsoon Woo, swoo@fhcrc.org and Raphael Gottardo <raphael.gottardo@ircm.qc.ca>

### References

Xuekui Zhang, Gordon Robertson, Sangsoon Woo, Brad G. Hoffman, and Raphael Gottardo, "Probabilistic Inference for Nucleosome Positioning with MNase-based or Sonicated Short-read Data" PlosONE, under review.

### See Also

`PING`

`plotSummary`