clim kwarg in expressionPalette or expressionColors about gigascatter.jl

[gszep]

maybe with using the cgrad method? https://docs.juliaplots.org/latest/generated/colorschemes/

[gszep]

palette = cgrad(:viridis)
scale(x::Number,y::StepRangeLen) = ( x - minimum(y) )/( maximum(y) - minimum(y) )
color = map( x->palette[scale(x,channelRange)], fcsdata[!,channel] )

[exaexa]

Yes, this would help a lot. The scaling is now (partially) done by scaleMinMax (see README examples). Perhaps the easiest way to have that working would be to supply kwargs min/max to the scaleMinMax function that set the bounds properly. Would that work for you?

See https://github.com/LCSB-BioCore/GigaScatter.jl/blob/master/src/colors.jl#L114 for scaleMinMax.

I didn't know about cgrad, I guess I can replace a bit of the code with that, e.g. this https://github.com/LCSB-BioCore/GigaScatter.jl/blob/master/src/colors.jl#L28 .

[gszep]

I realised that calling palette(::Float64) is 10x slower than palette(::Int64) and therefore it makes sense to pre-compute color indexes for each event. Here is the snippet I am working with:

channelRange, nlevels = range(-2,7,length=50), 10
palette = cgrad(:viridis,nlevels,categorical=true)

channels = names(fcsdata)
channel = Observable(first(channels))

function toIndex(x::AbstractVector{<:Number})
    min,max = extrema(channelRange)
    scaled = ( x .- min ) / ( max - min )
    
    @. scaled[scaled<0] = 0
    @. scaled[1<scaled] = 1
    
    return @. trunc(Int,nlevels*scaled) + 1
end

colorIndex = combine(fcsdata,[ col => toIndex => col for col ∈ names(fcsdata) ] )
color = Observable(palette[ colorIndex[!,channel[]] ])

############################# update on channel change
on(channel) do channel
    color[] = palette[ colorIndex[!,channel] ]
end

here color[] is a Vector{RGBA{Float64}} which is a good input for the color kwarg in Makie.jl. Is it also suitable for GigaScatter.jl?

[exaexa]

the slowness is quite likely because the float version is doing the interpolation with all colors; I guess I can do the same with our ColorPalette. I'll likely get to it tomorrow.