RejectionBy: Given an object of candidate pairs, reject false positives.
In npcooley/SynExtend: Tools for Comparative Genomics

RejectionBy

R Documentation

Given an object of candidate pairs, reject false positives.

Description

This function is designed to work internally to functions within SynExtend so it works on relatively simple atomic vectors and has little overhead checking.

Usage

RejectionBy(input,
            criteria = list("fdr" = 1e-5,
                            "centroidthreshold" = list("globalpid" = 0.3),
                            "glmforbiddencolumns" = c("alitype"),
                            "lmforbiddencolumns" = c("response",
                                                     "alitype"),
                            "kargs" = list("max" = 15,
                                           "scalar" = 4,
                                           "unitnorm" = TRUE)),
            rankby = "rawscore",
            method = "direct",
            supportedcolumns = c("consensus",
                                 "kmerdist",
                                 "featurediff",
                                 "localpid",
                                 "globalpid",
                                 "matchcoverage",
                                 "localscore",
                                 "deltabackground",
                                 "rawscore",
                                 "response"),
            dropinappropriate = FALSE)

Arguments

`input`	A data.frame; currently set up to take in an internal representation of the columns eventually printed by `SummarizePairs`. Supported column names are enumerated in the `supportedcolumns` argument.
`criteria`	A list of named objects that control the nobs of various rejection routines.
`rankby`	A colname in the the `input` data.frame which will be used for the identification of false positives.
`method`	Character; identify a method by which to reject false positives, currently supported methods include: `glm`, `lm`, `kmeans`, and `direct`.
`supportedcolumns`	Character; a vector of column names to select internal variables for candidate pair evaluation.
`dropinappropriate`	Logical; `FALSE` by default. If `TRUE`, should a method imply that it is incapable of logically separating true positives from false positives, no candidates are returned. If `FALSE`, all candidates are returned. Only really applicable in cases where very few initial candidates are identified.

Details

RejectionBy is yet another attempt at building a logical but simple routine for rejecting false positive candidate pairs in a scenario where it is difficult to proxy a distribution of appropriate false-positives.