mporder: Order markers within linkage groups
In mpMap: Multi-parent RIL genetic analysis

Description Usage Arguments Details Value See Also

Orders markers within linkage groups using two-point or multipoint probabilities. Two-point ordering is based on estimated recombination fractions; multi-point ordering is based on R/qtl ripple function.

  mporder(object, chr, type = c("2", "m"),
    mapfx = c("haldane", "kosambi"), rm.rf = TRUE,
    window = 3, repeats = 1,
    criterion = c("Path_length", "AR_events", "AR_deviations", "Gradient_raw", "Inertia", "Least_squares", "minXO", "lkhdsum"),
    missfx = 2, ...)

`object`	Object of class `mpcross`
`chr`	Selected chromosomes or linkage groups to order
`type`	Which type of ordering to perform - two-point or multipoint
`rm.rf`	Flag for whether to remove recombination fraction values from 2-pt ordering which have missing values
`window`	Window size for multipoint ordering
`repeats`	Number of times to repeat multipoint ordering
`mapfx`	Map function to use to compute final cM positions
`criterion`	Criterion used in 2-pt ordering to determine best order
`missfx`	Function to use to fill missing recombination fractions. See `fill`
`...`	Additional arguments

Two-point ordering
To use the two-point ordering, the recombination fractions between all pairs of markers must first be estimated. If there are missing values in this matrix, the markers with the largest number of missing values will be removed until there are no missing values left. These markers will not be used in the ordering and are recommended to be inserted into the resulting framework map using add3pt later.

Multiple methods are used to investigate optimal two-point orderings. These are taken from the package seriation and include simulated annealing, hierarchical clustering, and traveling salesman solver. The orders are compared on the basis of the argument criterion. Thus the total path length, or sum of adjacent recombination fractions can be minimized; or the number of Anti-Robinson events/deviations; or the number of crossovers; or the sum of the adjacent two-point LOD scores.

Multi-point ordering
The multi-point ordering assumes that there is a pre-existing map, and then repeatedly applies the ripple function in R/qtl to investigate local permutations of the order. These orderings are constrained by the arguments window and repeats, which determine how large the perturbations are and how many are considered. Large values of window are very time consuming; recommended values are 5 or less, due to the number of permutations which must be considered. Large values of repeats will eventually converge to an ordering in which all local rearrangements of size window have been optimized with respect to the number of crossovers.

The original object with a new map component. Any pre-existing map will be retained as component $oldmap.

mpestrf, mpgroup, add3pt, seriate, ripple