Add tutorial: Replicate Quality Control

[HLWeil]

Summary

I would like to add a tutorial about how to perform a quality control of sample replicates using:

1. Deedle for reading a frame containing the data
2. & 3. FSharp.Stats to impute missing values and cluster the samples
3. CyJS.NET to visualize the results

This tutorial is meant as a kind of protocol. @kMutagene should this go into the advanced or the data science category?

[kMutagene]

@HLWeil I think this fits better into the advanced category.

[bvenn]

I don't know what you are planning to do but I think there are 3 common ways to determine the sample replicability:

• histograms of coefficient of variation of the replicate groups
• column wise hierarchical clustering with subsequent analysis of cluster tree purity
• column wise PCA

I'm curious about your visualisation using CyJS.NET 🚀

Update: Sorry, it was not planned to close the issue 😅

[HLWeil]

Further improve tutorial quality:

Code quality

let imputedFrame = 
    let rowKeyMap = rawFrame.RowKeys |> Seq.indexed |> Map.ofSeq
    let columnKeyMap = rawFrame.ColumnKeys |> Seq.indexed |> Map.ofSeq
    Frame.ofJaggedArray imputedData
    |> Frame.mapRowKeys (fun r -> rowKeyMap.[r])
    |> Frame.mapColKeys (fun c -> columnKeyMap.[c])

should be replaced with

let imputedFrame = 
    Frame.ofJaggedArray imputedData
    |> Frame.indexRowsWith rawFrame.RowKeys
    |> Frame.indexColsWith rawFrame.ColumnKeys

---

might this

// Function for flattening the cluster tree to an edgelist
let hClustToEdgeList (f : int -> 'T) (hClust : HierarchicalClustering.Cluster<'T>) =
    let rec loop (d,nodeLabel) cluster=
        match cluster with
        | HierarchicalClustering.Node (id,dist,_,c1,c2) ->
            let t = f id
            loop (dist,t) c1
            |> List.append (loop (dist,t) c2)
            |> List.append [nodeLabel,t,d] 
        | HierarchicalClustering.Leaf (_,_,label)-> [(nodeLabel,label,d)]
    loop (0., f 0) hClust

or something similar be something for FSharp.Stats, @bvenn?

Spelling errors

• And it's not the best company, as missing values missing values
• Details about this can be found 003clusteringhierarchical.html and (Check index and correct link)
• Missing data is an constant companion of many data scientists