feat: Write metrics tables as outputs (#5)

* feat: Write metrics tables as outputs

Signed-off-by: jvfe <[email protected]>

* style: Black files

Signed-off-by: jvfe <[email protected]>

---------

Signed-off-by: jvfe <[email protected]>

microview/cli.py

@@ -70,9 +70,11 @@ def main(taxonomy: Path, csv_file: Path, output: Path) -> None:
             tax_results = get_tax_data(reports)
             # TODO: Improve this double check
             if parsed_result is not None:
-                tax_plots = generate_taxo_plots(tax_results, parsed_result["dataframe"])
+                tax_plots = generate_taxo_plots(
+                    tax_results, parsed_result["dataframe"], output_path=output
+                )
             else:
-                tax_plots = generate_taxo_plots(tax_results)
+                tax_plots = generate_taxo_plots(tax_results, output_path=output)
             render_base(tax_plots=tax_plots, dir_path=data_source, output_path=output)
         console.print(f"\n Done!\n", style="bold green")
     except Exception:

microview/plotting.py

@@ -29,6 +29,17 @@ def export_to_html(fig: Figure, div_id: str) -> str:
     )
 
 
+def write_table(df, output_path, path):
+    """
+    Write a dataframe to a file
+    """
+    dirpath = Path(output_path).parent.resolve() / Path("microview_tables")
+    Path(dirpath).mkdir(exist_ok=True)
+    path = dirpath / path
+
+    df.to_csv(path, sep="\t", index=False)
+
+
 def merge_with_contrasts(df, contrast_df, left_colname: Optional[str] = "index"):
     """
     Merges a dataframe with the dataframe containing contrasts (or groups)
@@ -40,10 +51,12 @@ def merge_with_contrasts(df, contrast_df, left_colname: Optional[str] = "index")
     return merged_df
 
 
-def plot_common_taxas(common_taxas_df, **kwargs):
+def plot_common_taxas(common_taxas_df, output_path, **kwargs):
     """
     Generate bar plot with most common taxas
     """
+    write_table(common_taxas_df, output_path, "common_taxas.tsv")
+
     return bar(
         common_taxas_df.sort_values(by=["value", "variable"], ascending=[False, True]),
         x="index",
@@ -60,10 +73,12 @@ def plot_common_taxas(common_taxas_df, **kwargs):
     )
 
 
-def plot_abund_div(abund_div_df, **kwargs):
+def plot_abund_div(abund_div_df, output_path, **kwargs):
     """
     Generate scatter plot of Pielou's Evenness and Shannon's Diversity (alpha)
     """
+    write_table(abund_div_df, output_path, "abund_diversity.tsv")
+
     return scatter(
         abund_div_df,
         x="Pielou Evenness",
@@ -77,10 +92,12 @@ def plot_abund_div(abund_div_df, **kwargs):
     )
 
 
-def plot_beta_pcoa(beta_pcoa, **kwargs):
+def plot_beta_pcoa(beta_pcoa, output_path, **kwargs):
     """
     Generate scatter plot of two first coordinates of Beta Diversity PCoA
     """
+    write_table(beta_pcoa, output_path, "beta_pcoa.tsv")
+
     fig = scatter(
         beta_pcoa,
         x="PC1",
@@ -94,7 +111,7 @@ def plot_beta_pcoa(beta_pcoa, **kwargs):
     return fig
 
 
-def generate_taxo_plots(tax_data: Dict, contrast_df=None) -> Dict:
+def generate_taxo_plots(tax_data: Dict, contrast_df=None, output_path=None) -> Dict:
     """
     Get all taxonomy plots
 
@@ -121,6 +138,7 @@ def generate_taxo_plots(tax_data: Dict, contrast_df=None) -> Dict:
         },
         template="plotly_white",
     )
+    write_table(tax_data["sample n reads"], output_path, "classified_reads.tsv")
 
     assigned.update_layout(
         xaxis={"categoryorder": "category ascending"},
@@ -146,6 +164,8 @@ def generate_taxo_plots(tax_data: Dict, contrast_df=None) -> Dict:
             .rename(columns={"index": "PC"})
         )
 
+        write_table(var_explained, output_path, "pcoa_variance_explained.tsv")
+
         pcoa_var = line(
             var_explained,
             x="PC",
@@ -162,25 +182,29 @@ def generate_taxo_plots(tax_data: Dict, contrast_df=None) -> Dict:
         ]
         merged_taxas_df = merge_with_contrasts(tax_data["common taxas"], contrast_df)
 
-        common_taxas = plot_common_taxas(merged_taxas_df, facet_col="group")
+        common_taxas = plot_common_taxas(
+            merged_taxas_df, output_path, facet_col="group"
+        )
         common_taxas.update_xaxes(matches=None)
 
         abund_div = plot_abund_div(
             merge_with_contrasts(tax_data["abund and div"], contrast_df),
+            output_path,
             color="group",
         )
         if plot_beta_div:
             betadiv_pcoa = plot_beta_pcoa(
                 merge_with_contrasts(pcoa_embed, contrast_df, left_colname="sample"),
+                output_path,
                 color="group",
             )
 
     else:
-        common_taxas = plot_common_taxas(tax_data["common taxas"])
+        common_taxas = plot_common_taxas(tax_data["common taxas"], output_path)
 
-        abund_div = plot_abund_div(tax_data["abund and div"])
+        abund_div = plot_abund_div(tax_data["abund and div"], output_path)
         if plot_beta_div:
-            betadiv_pcoa = plot_beta_pcoa(pcoa_embed)
+            betadiv_pcoa = plot_beta_pcoa(pcoa_embed, output_path)
 
     common_taxas.update_traces(showlegend=False)
     common_taxas.update_layout(

tests/test_data/microview_tables/abund_diversity.tsv

@@ -0,0 +1,5 @@
+index	Shannon Diversity	N Taxas	Pielou Evenness
+kaiju_test_2.txt	1.954647932741941	5	1.2144910453772946
+kaiju_test.txt	1.954647932741941	5	1.2144910453772946
+centrifuge_test.txt	0.029267352029729493	3	0.026640291876956372
+kraken_test.txt	3.0042033815854	181	0.5778984509401328

tests/test_data/microview_tables/beta_pcoa.tsv

@@ -0,0 +1,5 @@
+sample	PC1	PC2
+kaiju_test_2.txt	-0.25049691655478007	-0.04160014143773811
+kaiju_test.txt	-0.25049691655478046	-0.04160014143773799
+centrifuge_test.txt	0.7486472813642113	-0.00023745578005218948
+kraken_test.txt	-0.24765344825465063	0.08343773865552835

tests/test_data/microview_tables/classified_reads.tsv

@@ -0,0 +1,9 @@
+index	variable	value
+kaiju_test_2.txt	assigned	100.0
+kaiju_test.txt	assigned	100.0
+centrifuge_test.txt	assigned	0.2763703362505758
+kraken_test.txt	assigned	100.0
+kaiju_test_2.txt	unassigned	0.0
+kaiju_test.txt	unassigned	0.0
+centrifuge_test.txt	unassigned	99.72362966374942
+kraken_test.txt	unassigned	0.0

tests/test_data/microview_tables/common_taxas.tsv

@@ -0,0 +1,37 @@
+index	variable	value
+kraken_test.txt	unclassified	
+kraken_test.txt	root	
+kraken_test.txt	other	21.488
+kraken_test.txt	Ustilaginoidea virens nonsegmented virus 2	9.13
+kraken_test.txt	Tomato leaf curl alphasatellite	5.64
+kraken_test.txt	Siphoviridae	
+kraken_test.txt	Gordonia phage GTE6	38.56
+kraken_test.txt	Cymbidium mosaic virus	15.54
+kraken_test.txt	Circovirus-like genome SAR-B	9.64
+kaiju_test_2.txt	unclassified	
+kaiju_test_2.txt	root	
+kaiju_test_2.txt	other	0.0
+kaiju_test_2.txt	Ustilaginoidea virens nonsegmented virus 2	11.7
+kaiju_test_2.txt	Tomato leaf curl alphasatellite	6.42
+kaiju_test_2.txt	Siphoviridae	
+kaiju_test_2.txt	Gordonia phage GTE6	49.61
+kaiju_test_2.txt	Cymbidium mosaic virus	19.92
+kaiju_test_2.txt	Circovirus-like genome SAR-B	12.35
+kaiju_test.txt	unclassified	
+kaiju_test.txt	root	
+kaiju_test.txt	other	0.0
+kaiju_test.txt	Ustilaginoidea virens nonsegmented virus 2	11.7
+kaiju_test.txt	Tomato leaf curl alphasatellite	6.42
+kaiju_test.txt	Siphoviridae	
+kaiju_test.txt	Gordonia phage GTE6	49.61
+kaiju_test.txt	Cymbidium mosaic virus	19.92
+kaiju_test.txt	Circovirus-like genome SAR-B	12.35
+centrifuge_test.txt	unclassified	99.72
+centrifuge_test.txt	root	0.05
+centrifuge_test.txt	other	0.0
+centrifuge_test.txt	Ustilaginoidea virens nonsegmented virus 2	
+centrifuge_test.txt	Tomato leaf curl alphasatellite	
+centrifuge_test.txt	Siphoviridae	0.23
+centrifuge_test.txt	Gordonia phage GTE6	
+centrifuge_test.txt	Cymbidium mosaic virus	
+centrifuge_test.txt	Circovirus-like genome SAR-B	

tests/test_data/microview_tables/pcoa_variance_explained.tsv

@@ -0,0 +1,5 @@
+PC	Variance Explained
+PC1	0.9862442839054677
+PC2	0.013755716094532197
+PC3	0.0
+PC4	0.0