More olfaction plots

electrolyte_fm/data_modules/mixture_dataset.py

@@ -68,7 +68,7 @@ class ComponentDataModule(LightningDataModule):
     def __init__(
         self,
         path: str,
-        target_columns: Union[str, List[str]],
+        target_columns: list | str,
         n_components: int = 2,
         tokenizer: Optional[str] = None,
         batch_size: int = 64,

opt/design/olfaction_trends.jl

@@ -59,16 +59,13 @@ function plot_scent_by_type(df::DataFrame)
     types = sort(unique(df.type))
     scent_cols = get_scent_cols(df)
     colors = MISTStyle.CAT_COLORS
-    linestyles = [
-        :solid, (:dash, :dense), (:dot, :dense),
-        :dashdot, :dashdotdot, (:dot, :loose), (:dash, :loose)
-    ]
+    markers = [:circle, :rect, :diamond, :utriangle, :dtriangle, :pentagon, :cross, :xcross]
 
     scent_colors = Dict(s => colors[mod1(i, length(colors))] for (i, s) in enumerate(scent_cols))
-    scent_linestyles = Dict(s => linestyles[mod1(i, length(linestyles))] for (i, s) in enumerate(scent_cols))
+    scent_markers = Dict(s => markers[mod1(i, length(markers))] for (i, s) in enumerate(scent_cols))
 
-    fig = Figure(size=(250mm, 120mm), figure_padding=(3, 3, 3, 3))
-    n_cols = ceil(Int, length(types) / 2)
+    fig = Figure(size=(175mm, 170mm), figure_padding=(3, 3, 3, 3))
+    n_cols = ceil(Int, length(types) / 4)
 
     for (i, type) in enumerate(types)
         df_type = filter(row -> row.type == type, df)
@@ -99,17 +96,18 @@ function plot_scent_by_type(df::DataFrame)
                     sigmoid.(y_mean .+ y_std),
                     color=(scent_colors[scent], 0.2)
                 )
-                lines!(
+                scatterlines!(
                     ax, df_type.n_carbon, y_prob,
                     color=scent_colors[scent],
-                    linestyle=scent_linestyles[scent]
+                    marker=scent_markers[scent],
+                    markersize=5
                 )
 
                 push!(
                     plotted_elements,
-                    LineElement(
+                    MarkerElement(
                         color=scent_colors[scent],
-                        linestyle=scent_linestyles[scent]
+                        marker=scent_markers[scent],
                     )
                 )
                 push!(plotted_labels, scent)
@@ -139,16 +137,18 @@ function plot_branched_scent_comparison(df::DataFrame)
     max_variants = maximum(nrow(filter(row -> row.type == type, df)) for type in types)
     scent_colors = [colors[mod1(i, length(colors))] for i in eachindex(scents_to_plot)]
 
-    fig = Figure(size=(175mm, 140mm), figure_padding=(5, 5, 5, 5))
+    fig = Figure(size=(175mm, 120mm), figure_padding=(5, 5, 5, 5))
+    gl = GridLayout(fig[1, 1])
 
     for (i, type) in enumerate(types)
         df_type = filter(row -> row.type == type, df)
         row_pos, col_pos = grid_position(i, 2)
 
         ax = Axis(
-            fig[row_pos, col_pos],
+            gl[row_pos, col_pos],
             ylabel="Probability",
             title=type,
+            titlesize=6,
             xticklabelrotation=π/3,
             xticklabelalign=(:right, :center),
             limits=(0.5, max_variants + 0.5, 0, 1.3),
@@ -180,8 +180,14 @@ function plot_branched_scent_comparison(df::DataFrame)
         end
     end
 
+    sublabels = ["a", "b", "c", "d", "e", "f"]
+    for (i, label) in enumerate(sublabels[1:length(types)])
+        row_pos, col_pos = grid_position(i, 2)
+        sublabel!(gl[row_pos, col_pos, TopLeft()], label; left=15pt)
+    end
+
     Legend(
-        fig[3, :],
+        fig[2, 1],
         [PolyElement(color=c) for c in scent_colors],
         scents_to_plot,
         orientation=:horizontal,

opt/design/src/hydrocarbons.jl

@@ -22,25 +22,27 @@ tert_alkyl(n::Int) = n < 4 ? "C"^n : "C(C)(C)" * "C"^(n - 3)
 
 function branched_esters(n_total::Int)
     esters = []
-    # Generate esters with different chain length splits
     for n_acyl in 1:(n_total - 1)
         n_alkoxy = n_total - n_acyl
-        # Linear acyl, iso alkoxy (if n_alkoxy >= 3)
+
+        push!(esters, (; smi="O=C(O" * "C"^n_alkoxy * ")" * "C"^(n_acyl - 1),
+                       branch_pattern="linear"))
+
         if n_alkoxy >= 3
             push!(esters, (; smi="O=C(O" * iso_alkyl(n_alkoxy) * ")" * "C"^(n_acyl - 1),
                            branch_pattern="iso-alkoxy"))
         end
-        # Iso acyl, linear alkoxy (n_acyl - 1 >= 3, so n_acyl >= 4)
+
         if n_acyl >= 4
             push!(esters, (; smi="O=C(O" * "C"^n_alkoxy * ")" * iso_alkyl(n_acyl - 1),
                            branch_pattern="iso-acyl"))
         end
-        # Tert acyl, linear alkoxy (n_acyl - 1 >= 4, so n_acyl >= 5)
+
         if n_acyl >= 5
             push!(esters, (; smi="O=C(O" * "C"^n_alkoxy * ")" * tert_alkyl(n_acyl - 1),
                            branch_pattern="tert-acyl"))
         end
-        # Both iso (n_acyl - 1 >= 3 and n_alkoxy >= 3, so n_acyl >= 4 and n_alkoxy >= 3)
+
         if n_acyl >= 4 && n_alkoxy >= 3
             push!(esters, (; smi="O=C(O" * iso_alkyl(n_alkoxy) * ")" * iso_alkyl(n_acyl - 1),
                            branch_pattern="iso-both"))
@@ -51,30 +53,32 @@ end
 
 function branched_ethers(n_total::Int)
     ethers = []
-    # Generate ethers with different chain length splits
     for n_left in 1:(n_total - 1)
         n_right = n_total - n_left
-        # Linear left, iso right (if n_right >= 3)
+
+        push!(ethers, (; smi="C"^n_left * "O" * "C"^n_right,
+                       branch_pattern="linear"))
+
         if n_right >= 3
             push!(ethers, (; smi="C"^n_left * "O" * iso_alkyl(n_right),
                            branch_pattern="iso-right"))
         end
-        # Iso left, linear right (if n_left >= 3)
+
         if n_left >= 3
             push!(ethers, (; smi=iso_alkyl(n_left) * "O" * "C"^n_right,
                            branch_pattern="iso-left"))
         end
-        # Tert left, linear right (if n_left >= 4)
+
         if n_left >= 4
             push!(ethers, (; smi=tert_alkyl(n_left) * "O" * "C"^n_right,
                            branch_pattern="tert-left"))
         end
-        # Linear left, tert right (if n_right >= 4)
+
         if n_right >= 4
             push!(ethers, (; smi="C"^n_left * "O" * tert_alkyl(n_right),
                            branch_pattern="tert-right"))
         end
-        # Both iso (if both sides >= 3)
+
         if n_left >= 3 && n_right >= 3
             push!(ethers, (; smi=iso_alkyl(n_left) * "O" * iso_alkyl(n_right),
                            branch_pattern="iso-both"))
@@ -85,29 +89,30 @@ end
 
 function branched_alcohols(n_total::Int)
     alcohols = []
-    # Secondary alcohols - OH on carbon with iso branching (if n >= 3)
+
+    push!(alcohols, (; smi="O" * "C"^n_total,
+                     branch_pattern="primary-linear"))
+
     if n_total >= 3
         push!(alcohols, (; smi="O" * iso_alkyl(n_total),
                          branch_pattern="secondary-at-O-iso"))
     end
-    # Tertiary alcohols - OH on carbon with tert branching (if n >= 4)
+
     if n_total >= 4
         push!(alcohols, (; smi="O" * tert_alkyl(n_total),
                          branch_pattern="tertiary-at-O-tert"))
     end
-    # Secondary alcohols (OH on internal carbon, linear)
+
     for pos in 2:(n_total - 1)
-        # Linear secondary alcohol
         push!(alcohols, (; smi="C"^(pos - 1) * "C(O)" * "C"^(n_total - pos),
                          branch_pattern="secondary-linear"))
 
-        # Tertiary alcohol with branching at OH position (both sides must exist)
         if pos >= 2 && pos <= n_total - 2
             push!(alcohols, (; smi="C"^(pos - 1) * "C(O)(C)" * "C"^(n_total - pos - 1),
                              branch_pattern="tertiary-at-O"))
         end
     end
-    # Tertiary alcohols (3 distinct alkyl groups)
+
     if n_total >= 4
         for n_side1 in 1:(n_total - 3)
             for n_side2 in 1:(n_total - n_side1 - 2)
@@ -122,29 +127,30 @@ end
 
 function branched_thiols(n_total::Int)
     thiols = []
-    # Secondary thiols - SH on carbon with iso branching (if n >= 3)
+
+    push!(thiols, (; smi="S" * "C"^n_total,
+                   branch_pattern="primary-linear"))
+
     if n_total >= 3
         push!(thiols, (; smi="S" * iso_alkyl(n_total),
                        branch_pattern="secondary-at-S-iso"))
     end
-    # Tertiary thiols - SH on carbon with tert branching (if n >= 4)
+
     if n_total >= 4
         push!(thiols, (; smi="S" * tert_alkyl(n_total),
                        branch_pattern="tertiary-at-S-tert"))
     end
-    # Secondary thiols (SH on internal carbon, linear)
+
     for pos in 2:(n_total - 1)
-        # Linear secondary thiol
         push!(thiols, (; smi="C"^(pos - 1) * "C(S)" * "C"^(n_total - pos),
                        branch_pattern="secondary-linear"))
 
-        # Tertiary thiol with branching at SH position (both sides must exist)
         if pos >= 2 && pos <= n_total - 2
             push!(thiols, (; smi="C"^(pos - 1) * "C(S)(C)" * "C"^(n_total - pos - 1),
                            branch_pattern="tertiary-at-S"))
         end
     end
-    # Tertiary thiols (3 distinct alkyl groups)
+
     if n_total >= 4
         for n_side1 in 1:(n_total - 3)
             for n_side2 in 1:(n_total - n_side1 - 2)
@@ -205,15 +211,15 @@ end
 
 function fragrance_compounds(n::Int)
     df = DataFrame(vcat(
-        [(; type="Alkanes", smi=alkane(n)) for n in 1:n],
+        # [(; type="Alkanes", smi=alkane(n)) for n in 1:n],
         [(; type="Esters", smi=ester(n)) for n in 1:n],
         [(; type="Ethers", smi=ether(n)) for n in 3:n],
         [(; type="Alcohols", smi=alcohol(n)) for n in 1:n],
         [(; type="Aldehydes", smi=aldehyde(n)) for n in 1:n],
         [(; type="Carboxylic acids", smi=carboxylic_acid(n)) for n in 2:n],
-        [(; type="Alkenes", smi=alkene(n)) for n in 2:n],
-        [(; type="Alkynes", smi=alkyne(n)) for n in 2:n],
-        [(; type="Arenes", smi=arene(n)) for n in 0:n],
+        # [(; type="Alkenes", smi=alkene(n)) for n in 2:n],
+        # [(; type="Alkynes", smi=alkyne(n)) for n in 2:n],
+        # [(; type="Arenes", smi=arene(n)) for n in 0:n],
         [(; type="Thiol", smi=thiol(n)) for n in 1:n],
 
     ))
@@ -223,10 +229,10 @@ end
 
 function branched_fragrance_compounds(n::Int)
     df = DataFrame(vcat(
-        [(; type="Branched Esters", row...) for row in branched_esters(n)],
-        [(; type="Branched Ethers", row...) for row in branched_ethers(n)],
-        [(; type="Branched Alcohols", row...) for row in branched_alcohols(n)],
-        [(; type="Branched Thiols", row...) for row in branched_thiols(n)],
+        [(; type="Esters", row...) for row in branched_esters(n)],
+        [(; type="Ethers", row...) for row in branched_ethers(n)],
+        [(; type="Alcohols", row...) for row in branched_alcohols(n)],
+        [(; type="Thiols", row...) for row in branched_thiols(n)],
     ))
     n_carbon!(df)
     return df