ext/VisualizationExt.jl: Make CairoMakie optional and plot_diagnostic…

…s as extension.

Project.toml

@@ -11,6 +11,12 @@ Clustering = "aaaa29a8-35af-508c-8bc3-b662a17a0fe5"
 DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
 ProgressMeter = "92933f4c-e287-5a05-a399-4b506db050ca"
 
+[weakdeps]
+CairoMakie = "13f3f980-e62b-5c42-98c6-ff1f3baf88f0"
+
+[extensions]
+VisualizationExt = "CairoMakie"
+
 [compat]
 BioSequences = "3.1.6"
 CSV = "0.10.14"
@@ -21,7 +27,8 @@ ProgressMeter = "1.10.2"
 julia = "1.10"
 
 [extras]
+CairoMakie = "13f3f980-e62b-5c42-98c6-ff1f3baf88f0"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["Test"]
+test = ["Test", "CairoMakie"]

ext/VisualizationExt.jl

@@ -0,0 +1,82 @@
+module VisualizationExt
+    using LineageCollapse
+    using CairoMakie
+    using DataFrames
+    export plot_diagnostics
+
+    """
+        plot_diagnostics(df::DataFrame)
+
+    Generate professional diagnostic plots for the processed data using CairoMakie.
+
+    # Arguments
+    - `df::DataFrame`: Processed DataFrame with lineage information.
+
+    # Returns
+    - `CairoMakie.Figure`: A composite figure with multiple diagnostic visualizations.
+    """
+    function plot_diagnostics(df::DataFrame)
+        fig = Figure(size=(1200, 1000), fontsize=12)
+
+        # Cluster Size Distribution
+        if hasproperty(df, :cluster_size)
+            ax1 = Axis(fig[1, 1], title="Cluster Size Distribution", xlabel="Cluster Size", ylabel="Frequency")
+            hist!(ax1, df.cluster_size, bins=50, color=:skyblue, strokecolor=:white, strokewidth=1)
+        else
+            ax1 = Axis(fig[1, 1], title="Cluster Size Distribution Not Available")
+        end
+
+        # CDR3 Length vs Cluster Size
+        if hasproperty(df, :cdr3_length) && hasproperty(df, :cluster_size)
+            ax2 = Axis(fig[1, 2], title="CDR3 Length vs Cluster Size", xlabel="CDR3 Length", ylabel="Cluster Size")
+            scatter!(ax2, df.cdr3_length, df.cluster_size, color=:darkblue, markersize=4, alpha=0.5)
+        else
+            ax2 = Axis(fig[1, 2], title="CDR3 Length vs Cluster Size Not Available")
+        end
+
+        # CDR3 Frequency Distribution
+        if hasproperty(df, :cdr3_frequency)
+            ax3 = Axis(fig[2, 1], title="CDR3 Frequency Distribution", xlabel="CDR3 Frequency", ylabel="Count")
+            hist!(ax3, df.cdr3_frequency, bins=50, color=:lightgreen, strokecolor=:white, strokewidth=1)
+        else
+            ax3 = Axis(fig[2, 1], title="CDR3 Frequency Distribution Not Available")
+        end
+
+        # Top 10 V Genes
+        ax4 = Axis(fig[2, 2], title="Top 10 V Genes", xlabel="V Gene", ylabel="Count")
+        if hasproperty(df, :v_call_first)
+            try
+                v_gene_counts = sort(combine(groupby(df, :v_call_first), nrow => :count), :count, rev=true)
+                if nrow(v_gene_counts) > 10
+                    v_gene_counts = v_gene_counts[1:10, :]
+                end
+                barplot!(ax4, v_gene_counts.count, color=:orange)
+                ax4.xticks = (1:nrow(v_gene_counts), v_gene_counts.v_call_first)
+                ax4.xticklabelrotation = π/3
+            catch e
+                @warn "Error processing V gene counts: $e"
+                ax4.title = "Top 10 V Genes (Error in Processing)"
+            end
+        else
+            ax4.title = "Top 10 V Genes Not Available"
+        end
+
+        # Rotate x-axis labels for all plots
+        for ax in [ax1, ax2, ax3, ax4]
+            ax.xticklabelrotation = π/4
+        end
+
+        # Add a title to the entire figure
+        Label(fig[0, :], "Diagnostic Plots for Lineage Collapse", fontsize=20)
+
+        # Adjust layout
+        for (label, layout) in zip(["A", "B", "C", "D"], [fig[1,1], fig[1,2], fig[2,1], fig[2,2]])
+            Label(layout[1, 1, TopLeft()], label,
+                fontsize = 26,
+                padding = (6, 6, 6, 6),
+                halign = :right)
+        end
+
+        return fig
+    end
+end

src/LineageCollapse.jl

@@ -4,12 +4,11 @@ module LineageCollapse
     using ProgressMeter
     using Clustering
     using BioSequences
-    using CairoMakie
 
     export load_data, preprocess_data, process_lineages, plot_diagnostics
+    function plot_diagnostics(df::DataFrame) end
 
     include("data_loading.jl")
     include("preprocessing.jl")
     include("lineage_processing.jl")
-    include("visualization.jl")
 end

src/lineage_processing.jl

@@ -55,7 +55,9 @@ function process_lineages(df::DataFrame;
     grouped = groupby(df, [:v_call_first, :j_call_first, :cdr3_length])
     processed_groups = Vector{DataFrame}()
 
-    @showprogress "Processing lineages" for group in grouped
+    prog = Progress(length(grouped), desc="Processing lineages")
+    for group in grouped
+        next!(prog)
         if nrow(group) > 1
             dist = pairwise_hamming(LongDNA{4}.(group.cdr3))
             hclusters = hclust(dist, linkage=:average)
@@ -76,6 +78,6 @@ function process_lineages(df::DataFrame;
             push!(processed_groups, cgroup)
         end
     end
-
+    finish!(prog)
     return vcat(processed_groups...)
 end

test/runtests.jl

@@ -1,9 +1,9 @@
 using Test
 using DataFrames
 using BioSequences
+using CairoMakie
 using LineageCollapse
 using CSV
-using CairoMakie
 
 @testset "LineageCollapse.jl" begin
     @testset "Data Loading" begin
@@ -88,6 +88,7 @@ using CairoMakie
         end
     end
 
+    # Test visualization functionality only if CairoMakie is available
     @testset "Visualization" begin
         test_df = DataFrame(
             cluster_size = [1, 2, 3, 4, 5],