struct Experiment
    runs::Int
    positive_controls::Int
    negative_controls::Int
end 

Collect the number of runs and controls in an experiment.

struct PlateArray
    wells::BitMatrix
    positives::BitMatrix 
    negatives::BitMatrix
end

A PlateArray object describes the layout of a microwell plate that includes the active experimental wells and controls.

objectives

Predefined objectives for the place_controls algorithm

solvers

Predefined solvers for the place_controls algorithm

size(p::PlateArray)

Return the size (rows x colummns) of the BitMatrices that define the PlateArray

function MILP(P::Int,N::Int,wells::BitMatrix;objective::Function=hybrid,minimize=true,timelimit=100)

MILP solver for control placment. Requires Gurobi licence.

Arguments

• wells: A BitMatrix indicating the shape and active wells, use trues(n,m) for a full n x m plate.
• P: The integer number of positive controls
• N: The integer number of negative controls

Keyword Arguments

• objective: the objective type for the MILP solver. Must be either 'minimax' or 'hybrid'.
• timelimit: time limit in seconds for the solver to return a suboptimal solution if it hasn't found an optimal one
• minimize: if true, the solver minimizes the distance from experiment wells to control wells. if false, it will maximize (this is not useful for practical purposes but is helpful for assessing performance)

active_indices(plate::BitMatrix)

Compute the integer indices of active wells.

arrayer(wells::BitMatrix,experiments::Vararg{Experiment};kwargs...)

Array Experiment objects onto plates in three steps:

1. assign all experiments to as few plates as possible
2. partition plates that contain multiple experiments and select wells to hold each run. Use central wells first.
3. place a full complement of controls on each plate that has a given experiment

Arguments

• wells: A BitMatrix of active wells on each plate (block any inactive wells by setting them to false)
• experiments: Array a variable number of Experiment objects

Returns a Matrix of PlateArray objects with dimensions E x P, where E is the number of experimens and P the number of plates.

exchange(wells::BitMatrix,P::Int,N::Int;objective::Function=hybrid,minimize=true,restarts::Int=10,iterations::Int=1000,kwargs...)

Exchange solver for control placment

Arguments

• plate: A BitMatrix indicating the shape and active wells, use trues(n,m) for a full n x m plate.
• P: The integer number of positive controls
• N: The integer number of negative controls

Keyword Arguments

• objective : The objective function criteria, choose from minimax, LHS, and hybrid. The hybrid objective is slower than the other objectives because it calculates bounds to balance two objectives.
• minimize : flip the objective sign
• restarts : Number of solver restarts. The solver returns the best solution among the restarts
• iterations: number of exchange iterations per run.

function partition(wells::BitMatrix,expts::Vararg{Int})

Partition experiments onto a plate of active wells. Use center wells before using edge wells, and place the experiments into contiguous blocks.

Arguments

• wells a BitMatrix where each active well has a value of 'true'
• expts a Vararg Int indicating how many runs are present in each experiment

function place_controls(wells::BitMatrix,expt::Experiment;kwargs...)

Place optimal controls for detecting errors in microplate experiments

Arguments

• wells: A BitMatrix mask indicating the shape and active wells, use trues(n,m) for a full n x m plate.
• expt : An Experiment type indicating the number of runs as well as positive and negative controls

Keyword Arguments

• solver: The algorithm used to place the controls. There are currently two solvers available:
  1. exchange (default):
  2. MILP :
• objective: The objective the solver uses to score plate array candidates.
  1. minimax -> Minimize the maximum distance in wells from an active well to its nearest control.
  2. LHS -> Find an approximate latin hypercube sample of the available wells in the plate.
  3. hybrid (default) -> a weighted combination of both criteria

place_controls(wells::BitMatrix,P::Int,N::Int;solver::String="exchange",objective::String="hybrid",kwargs...)

Place optimal controls for detecting errors in microplate experiments

Arguments

• wells: A BitMatrix mask indicating the shape and active wells, use trues(n,m) for a full n x m plate.
• P: The integer number of positive controls
• N: The integer number of negative controls

Keyword Arguments

• solver: The algorithm used to place the controls. There are currently two solvers available:
  1. exchange (default):
  2. MILP :
• objective: The objective the solver uses to score plate array candidates.
  1. minimax -> Minimize the maximum distance in wells from an active well to its nearest control.
  2. LHS -> Find an approximate latin hypercube sample of the available wells in the plate.
  3. hybrid (default) -> a weighted combination of both criteria

runs(platearray::PlateArray)

Compute the non-control active wells of a PlateArray.