o 8iG @sdZddlZddlZddlZddlmZddlmZm Z m Z dede de dee fd d Z d ede dede de f d d Zd ede fddZd ede dede fddZddZedkr_edSdS)z Generate an HTML summary page for a variant that displays all plots in an easily-readable format. Usage: python generate_variant_summary_html.py N)Path)ListOptionalDict clustered_tsv variant_id model_datasetreturnc Cs|s td|gSz tj|dd}Wnty1}ztd|gWYd}~Sd}~ww||d|k}|jrGtd|dgSg}d |}||jvr||jd }t|rt | rd d t | d D} | D]} d| vr| dd } | r| | qs|s|jD]<} | d r|| jd }t|rt | rdd t | d D} | D]} d| vr| dd } | r| | qqtt|S)a Get prioritized models for a variant from clustered.tsv. Reads the models_prioritized_by_any-{model_dataset} column and parses the semicolon-separated format to extract model names. Parameters: ----------- clustered_tsv : Path Path to clustered.tsv file variant_id : str Variant ID to look up model_dataset : str Model dataset name (e.g., "Fetal Brain") Returns: -------- List[str] List of model names that prioritize this variant z'Warning: clustered.tsv does not exist:  )sepzError reading clustered.tsv: NrzWarning: Variant z not found in clustered.tsvzmodels_prioritized_by_any-rcSg|] }|r|qSstrip.0mr r *snakemake/generate_variant_summary_html.py <z3get_prioritized_models_from_tsv..;(cSr r rrr r rrJr)existsprintpdread_csv Exceptionemptycolumnsilocnotnastrrsplitappend startswithlistset) rrrdfe variant_rowmodelsaggregated_col models_str model_entriesentrymodelcolr r rget_prioritized_models_from_tsvsN       r1 variant_dircCsd|d|d|d|dgd}t|||}|D]}|d|d}|d|||d q|S) zFind all expected plot paths for a variant, regardless of whether they exist. Returns: dict with keys: 'barplot', 'barplot_superset', 'scatterplot', 'scatterplot_superset', 'profiles' (list) 01-model-specificity-barplot.png)01-model-specificity-barplot-superset.png02-model-scatterplot.html"02-model-scatterplot-superset.htmlbarplotbarplot_superset scatterplotscatterplot_supersetprofiles 03-profile-z.pngr<)pathr/r)r1r#r)r2rrrplotsprioritized_modelsr/ profile_pathr r rfind_expected_plotsTs  rBcCsddddgd}|d}|r||d<|d}|r ||d<|d}|r,||d<|d }|r8||d <t|d D]}|jd d }|d||dq?|S)zFind all plot files in the variant directory. Returns: dict with keys: 'barplot', 'barplot_superset', 'scatterplot', 'scatterplot_superset', 'profiles' (list) Nr7r3r8r4r9r5r:r6r;z03-profile-*.pngr=r<)r>r/)rsortedglobstemreplacer#)r2r? barplot_pathbarplot_superset_pathscatterplot_pathscatterplot_superset_path profile_file model_namer r r find_plotsps2  rN output_htmlr?c s|jdtdtffdd }d|d|d}|d7}|d 7}|d }|rIt|tr<|r<||}|d |d 7}n||}|d |d 7}n|d7}|d7}|d7}|d7}|d}|rt|trv|rv||}|d|d7}n||}|d|d7}n|d7}|d7}|d7}|d7}|d} | rt| tr| r|| }|d|d7}n|| }|d|d7}n|d7}|d7}|d7}|d7}|d} | rt| tr| r|| }|d|d7}n|| }|d|d7}n|d7}|d7}|d7}|d 7}|d!rT|d"7}|d!D]:} | d#} || }| d$d%} | r;|d&| d'd(|d)| d'd*7}q|d&| d'd(|d)| d'd+7}q|d,7}n|d-7}|d7}|d.7}|jjd/d/d0t|d1 }||Wd2n 1s~wYt d3|d2S)4z+Generate HTML summary page for the variant. file_pathr cs,ztj|WStyt|YSw)z/Get relative path from output HTML to the file.)osr>relpath ValueErrorr!)rP output_dirr rget_relative_paths   z(generate_html..get_relative_pathz Variant Summary: al ��

Variant Summary

z
z
z;

Model Specificity Barplot (Cluster-level)

r8z
Model prioritization across tissues/organs in the current cluster/model dataset
Model Specificity Barplot
a " alt="Model Specificity Barplot" onerror="this.style.display='none'; this.nextElementSibling.style.display='block';">
Plot will be available after generation
zG
No cluster-level barplot available
z z<

Model Specificity Barplot (Superset-level)

r9z
Model prioritization across tissues/organs in the superset (broader context)
Model Specificity Barplot (Superset)
a" alt="Model Specificity Barplot (Superset)" onerror="this.style.display='none'; this.nextElementSibling.style.display='block';">
Plot will be available after generation
zH
No superset-level barplot available
z3

Model Scatterplot (Cluster-level)

r:z
Interactive scatterplot showing variant scores across models in the current cluster/model dataset
a" frameborder="0" onerror="this.style.display='none'; this.nextElementSibling.style.display='block';">
Plot will be available after generation
zK
No cluster-level scatterplot available
z4

Model Scatterplot (Superset-level)

r;z
Interactive scatterplot showing variant scores across all models in the superset (broader context)