dropSimilarMarkers: Method to improve a genetic map.
In jtlovell/qtlTools: Data Processing and Plotting in Association with R/qtl

Description Usage Arguments Value Examples

dropSimilarMarkers finds markers that have a small recombination fraction and drops the one with combined greater segregation distortion and/or missing data. ***Note: if the cross object has many markers (>1000), avoid running this function on more than one chromosome at a time. Also make sure to run est.rf first and use re.est.map = FALSE***

dropSimilarMarkers(cross, chr = NULL, rf.threshold = 0.01,
  sd.weight = 1, na.weight = 1, keepEnds = FALSE, doNotDrop = NULL,
  verbose = TRUE, blockSize = 100, byChr = TRUE,
  runFullMatrix = FALSE, ...)

`cross`	The qtl cross object to search
`chr`	numeric, Should the analysis be restricted to a set of chromosomes. Defaults to all chromosomes in the cross object.
`rf.threshold`	The recombination fraction threshold to drop a marker. If est.rf has not been run on cross, it will be done so automatically. See qtl::est.rf for details
`sd.weight`	The weighting of segregation distortion rank in dropping a marker. Higher values relative to na.weight increase the weight of the sd rank. Setting a value of 0 removes sd as a factor in choosing the best marker.
`na.weight`	Same as sd.weight, but for the number of NAs.
`keepEnds`	Logical, should markers on the ends of the chromosomes always be retained?
`doNotDrop`	Character vector of markers to retain no matter their rfs.
`verbose`	Logical, should updates be printed?
`blockSize`	If not NULL, do an initial culling by splitting markers into blocks of this size. Smaller blocks run more quickly than large blocks, but when the total number of blocks excedes ~ 2000, it can take a very long time to parse the cross object into blocks.
`byChr`	Should the procedure be run chromosome-by-chromosome. If blocksize != NULL, this procedure is run following block-wise culling. If there are many thousands of markers it is recommended to run multiple block-wise calls prior to whole-chromosome procedures. In general, chromosomes with > 1k markers should first be culled using blockSize != NULL.
`runFullMatrix`	should the full matrix ever be assessed?
`...`	if recombination fractions are not included in the cross object, pass on additional arguments to est.rf.

A new cross object with the similar markers dropped.

set.seed(42)
map<-sim.map(len = c(50,20), n.mar = c(20,30), include.x=FALSE)
cross0<-sim.cross(map, n.ind=50, type="f2", map.function="kosambi",
   error.prob=.01, missing.prob = .05)
cross0<-est.rf(cross0)
cross1<-dropSimilarMarkers(cross0)
cross2<-dropSimilarMarkers(cross0, keepEnds=TRUE)
par(mfrow=c(2,1))
plot.map(cross0, cross1, main = "comparison of full and culled maps")
plot.map(cross0, cross2, main = "comparison of full and culled maps")