diff --git a/choppa/cli/cli.py b/choppa/cli/cli.py index 196df04..8aa6d7e 100644 --- a/choppa/cli/cli.py +++ b/choppa/cli/cli.py @@ -1,20 +1,133 @@ import click +from typing import Optional +from choppa.IO.input import FitnessFactory, ComplexFactory +from choppa.align.align import AlignFactory +from choppa.render.render import PublicationView, InteractiveView @click.group() @click.version_option() def cli(): - "Demo of click-app-template-repository" + "Integrated mutational and structural biology data into a concerted HTML view." -@cli.command(name="command") -@click.argument( - "example" +@cli.command(name="render", help="Create fitness view as a publication-ready PyMOL session file and a read-for-sharing interactive HTML file. ", short_help="Create fitness view as a publication-ready PyMOL session file and a read-for-sharing interactive HTML file.") +@click.option( + "-p", + "--pdb-file", + type=click.Path(exists=True, file_okay=True, dir_okay=False, writable=True), + help="Path to a PDB file to create fitness view for.", + required=True, +) +@click.option( + "-f", + "--fitness-file", + type=click.Path(exists=True, file_okay=True, dir_okay=False, writable=True), + help="Path to a CSV file with fitness data to create fitness view for.", + required=True, +) +@click.option( + "-op", + "--outfile-publication", + type=click.Path(exists=False, file_okay=True, dir_okay=False, writable=True), + help="Name of output file to write publication-ready PyMOL session file to. Should end in '.pse'", + required=True, +) +@click.option( + "-oi", + "--outfile-interactive", + type=click.Path(exists=False, file_okay=True, dir_okay=False, writable=True), + help="Name of output file to write ready-to-share interactive HTML file to. Should end in '.html'", + required=True, +) +@click.option( + "-f", + "--fitness-file", + type=click.Path(exists=True, file_okay=True, dir_okay=False, writable=True), + help="Path to a CSV file with fitness data to create fitness view for.", + required=True, +) +@click.option( + "-f", + "--fitness-file", + type=click.Path(exists=False, file_okay=True, dir_okay=False, writable=True), + help="Path to a CSV file with fitness data to create fitness view for.", + required=True, +) +@click.option( + "-f", + "--fitness-file", + type=click.Path(exists=False, file_okay=True, dir_okay=False, writable=True), + help="Path to a CSV file with fitness data to create fitness view for.", + required=True, +) +@click.option( + "-fv", + "--fitness-column", + type=click.STRING, + help="Name of the column in the fitness-file (-f/--fitness-file) that contains fitness values (e.g. LogEffect). If not defined, will default to 'fitness'.", + required=False, +) +@click.option( + "-ri", + "--residue-index-column", + type=click.STRING, + help="Name of the column in the fitness-file (-f/--fitness-file) that contains residue indices (e.g. 1, 2, .. n). If not defined, will default to 'residue_index'.", + required=False, +) +@click.option( + "-wt", + "--wildtype-column", + type=click.STRING, + help="Name of the column in the fitness-file (-f/--fitness-file) that contains wildtype residues (e.g. L, G, N). If not defined, will default to 'wildtype'.", + required=False, +) +@click.option( + "-mu", + "--mutant-column", + type=click.STRING, + help="Name of the column in the fitness-file (-f/--fitness-file) that contains mutant residues (e.g. L, G, N). If not defined, will default to 'mutant'.", + required=False, ) @click.option( - "-o", - "--option", - help="An example option", + "-c", + "--confidence-column", + type=click.STRING, + help="Name of the column in the fitness-file (-f/--fitness-file) that contains confidence values (e.g. counts). If not defined then LogoPlots in the HTML view will not display confidences.", + required=False, ) -def first_command(example, option): - "Command description goes here" - click.echo("Here is some output") \ No newline at end of file +def render( + pdb_file: Optional[str] = None, + fitness_file: Optional[str] = None, + outfile_publication: Optional[str] = None, + outfile_interactive: Optional[str] = None, + fitness_column : Optional[str] = None, + residue_index_column : Optional[str] = None, + wildtype_column : Optional[str] = None, + mutant_column : Optional[str] = None, + confidence_column : Optional[str] = None, + ): + + if not outfile_publication[-4:] == ".pse": + raise ValueError("--op/--outfile-publication should end in '.pse'.") + elif not outfile_interactive[-5:] == ".html": + raise ValueError("--oi/--outfile-interactive should end in '.html'.") + + fitness_dict = FitnessFactory(fitness_file, + confidence_colname=confidence_column + ).get_fitness_basedict() + complex = ComplexFactory(pdb_file).load_pdb() + complex_rdkit = ComplexFactory(pdb_file).load_pdb_rdkit() + + filled_aligned_fitness_dict = AlignFactory(fitness_dict, complex).align_fitness() + + PublicationView(filled_aligned_fitness_dict, + complex, + complex_rdkit, + fitness_threshold=0.7, + output_session_file=outfile_publication).render() + + InteractiveView(filled_aligned_fitness_dict, + complex, + complex_rdkit, + fitness_threshold=0.7, + output_session_file=outfile_interactive).render() \ No newline at end of file diff --git a/choppa/render/Template.html b/choppa/data/templates/Template.html similarity index 100% rename from choppa/render/Template.html rename to choppa/data/templates/Template.html diff --git a/choppa/data/templates/resources.py b/choppa/data/templates/resources.py new file mode 100644 index 0000000..0ffb1c8 --- /dev/null +++ b/choppa/data/templates/resources.py @@ -0,0 +1,6 @@ +import pkg_resources + +HTML_TEMPLATE = pkg_resources.resource_filename( + __name__, "Template.html" +) + diff --git a/choppa/render/.DS_Store b/choppa/render/.DS_Store deleted file mode 100644 index 35dc4cf..0000000 Binary files a/choppa/render/.DS_Store and /dev/null differ diff --git a/choppa/render/out.html b/choppa/render/out.html deleted file mode 100644 index 2064a95..0000000 --- a/choppa/render/out.html +++ /dev/null @@ -1,4251 +0,0 @@ - - - - - - - - - - - - -
-
-
- -
- -
- Protein residue surfaces are colored by mutability: -
-
- ⚪ : No amino acid substitutions tolerated - 🔴 : increasing tolerance for amino acid substitutions (increasing with 🔴 intensity) - 🟣 : No data for residue -
-
- -
- -
- Ligand-protein contacts are shown as dashed lines colored by: -
-
- ⬜ : Ligand contact is with amino acid side chain that has no tolerated substitutions - 🟩 : Ligand contact is with peptide backbone - 🟥 : Ligand contact is with amino acid side chain that has tolerated substitutions (increasing with 🔴 intensity) - 🟪 : No data for contacted residue -
-
- -
- -
- Fitness logo plots are shown on hover of residue atoms: -
-
- Top: Wildtype of this residue; this should correspond to the residue type that is being hovered over in the 3D view. - Left: amino acids at this position that are consistent with virus viability. Letter heights are scaled to indicate
the fractions of the viable viral populations with the particular residue at this position - Right: amino acids at this position that are present in the selected population at background frequencies,
and thus likely to be inconsistent with viral viability. Stop codons (*) can also be present in these populations
of unselected genomes - Coloring: Residue letters are either colored by type or by confidence if your input fitness data contains a confidence column.
The right-hand-side logoplot displays the colorbar for the range of confidences. -
-
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - \ No newline at end of file diff --git a/choppa/render/out.pse b/choppa/render/out.pse deleted file mode 100644 index 8532b98..0000000 Binary files a/choppa/render/out.pse and /dev/null differ diff --git a/choppa/render/render.py b/choppa/render/render.py index 42c3172..28d41fb 100644 --- a/choppa/render/render.py +++ b/choppa/render/render.py @@ -6,6 +6,7 @@ from choppa.render.utils import show_contacts, get_ligand_resnames_from_pdb_str, split_pdb_str, get_contacts_mda, biopython_to_mda from choppa.render.logoplots import LogoPlot, WHITE_EMPTY_SQUARE, render_singleres_logoplot +from choppa.data.templates.resources import HTML_TEMPLATE logging.basicConfig(stream=sys.stdout, level=logging.INFO) logger = logging.getLogger() @@ -414,7 +415,7 @@ def get_interaction_dict(self): return intn_dict - def inject_stuff_in_template(self, sdf_str, pdb_str, surface_coloring, logoplot_dict, template="Template.html", out_file="out.html"): + def inject_stuff_in_template(self, sdf_str, pdb_str, surface_coloring, logoplot_dict, template=HTML_TEMPLATE): """" Replaces parts of a template HTML with relevant bits of data to get to a HTML view of the (ligand-) protein, its fitness and its interactions (if any). @@ -452,7 +453,7 @@ def inject_stuff_in_template(self, sdf_str, pdb_str, surface_coloring, logoplot_ # add the PDB (protein) and SDF (ligand) with open(template, "rt") as fin: - with open(out_file, "wt") as fout: + with open(self.output_session_file, "wt") as fout: for line in fin: line = line.replace("{{PDB_INSERT}}", f"{pdb_str}") line = line.replace("{{SDF_INSERT}}", f"{sdf_str}")