chore: switch to click for argument parsing (#10)

Makefile

@@ -29,6 +29,6 @@ mypy:
 .PHONY: lint
 lint: flake8 isort-check black-check mypy
 
-.PHONY: pytest
-pytest:
+.PHONY: test
+test:
 	pytest .

chew/cli.py

@@ -0,0 +1,144 @@
+import typing
+
+import click
+
+from chew import __version__, compare, fingerprint, plot_compare, plot_var_het, stats
+
+
+@click.group()
+@click.version_option(__version__)
+@click.option("--verbose/--no-verbose", help="Enable verbose mode.", default=False)
+@click.pass_context
+def cli(ctx: click.Context, verbose: bool):
+    """Main entry point for CLI via click."""
+    ctx.ensure_object(dict)
+    ctx.obj["verbose"] = verbose
+
+
+@cli.command("fingerprint", help="Compute fingerprint to numpy .npz files.")
+@click.option("--min-coverage", type=int, default=5, help="Minimal required coverage.")
+@click.option("--reference", required=True, help="Path to reference FASTA file.")
+@click.option(
+    "--output-fingerprint",
+    required=True,
+    help="Path to output .npz file (extension will be added automatically if necessary)",
+)
+@click.option(
+    "--output-aafs/--no-output-aafs",
+    default=True,
+    help="Write alternate allele fractions to .npz file.",
+)
+@click.option("--input-bam", required=True, help="Path to input BAM file.")
+@click.option("--genome-release", required=False, default="GRCh37", help="Genome release used.")
+@click.option("--max-sites", type=int, default=0, help="Maximal number of sites to consider.")
+@click.option("--write-vcf/--no-write-vcf", default=False, help="Enable writing of call VCF.")
+@click.pass_context
+def cli_fingerprint(
+    ctx: click.Context,
+    min_coverage: int,
+    reference: str,
+    output_fingerprint: str,
+    output_aafs: bool,
+    input_bam: str,
+    genome_release: str,
+    max_sites: int,
+    write_vcf: bool,
+):
+    config = fingerprint.Config(
+        verbosity=2 if ctx.obj["verbose"] else 1,
+        min_coverage=min_coverage,
+        reference=reference,
+        output_fingerprint=output_fingerprint,
+        output_aafs=output_aafs,
+        input_bam=input_bam,
+        genome_release=genome_release,
+        max_sites=max_sites,
+        write_vcf=write_vcf,
+    )
+    fingerprint.run(config)
+
+
+@cli.command("compare", help="Perform fingeprint comparison.")
+@click.option(
+    "--output-prefix", type=str, default="chew-comparison", help="Path to comparison file."
+)
+@click.option("--min-mask-ones", type=int, help="Minimal number of ones in mask.")
+@click.option("--max-mask-ones", type=int, help="Maximal number of ones in mask.")
+@click.argument("fingerprints", nargs=-1)
+@click.pass_context
+def cli_compare(
+    ctx: click.Context,
+    output_prefix: str,
+    min_mask_ones: typing.Optional[int],
+    max_mask_ones: typing.Optional[int],
+    fingerprints: typing.List[str],
+):
+    config = compare.Config(
+        verbosity=2 if ctx.obj["verbose"] else 1,
+        output_prefix=output_prefix,
+        min_mask_ones=min_mask_ones,
+        max_mask_ones=max_mask_ones,
+        fingerprints=fingerprints,
+    )
+    compare.run(config)
+
+
+@cli.command("stats", help="Compute statistics from fingerprint .npz files.")
+@click.option("--output", default="chew-stats.txt", help="Path to output file.")
+@click.argument("fingerprints", nargs=-1)
+@click.pass_context
+def cli_stats(
+    ctx: click.Context,
+    output: str,
+    fingerprints: typing.List[str],
+):
+    config = stats.Config(
+        verbosity=2 if ctx.obj["verbose"] else 1,
+        output=output,
+        fingerprints=fingerprints,
+    )
+    stats.run(config)
+
+
+@cli.command("plot-compare", help="Plot result of 'ngs-chew compare'.")
+@click.option(
+    "--title", default="NGS Chew Comparison Plot", help="title to use for the output HTML file."
+)
+@click.argument("compare-out")
+@click.argument("out_html")
+@click.pass_context
+def cli_plot_compare(
+    ctx: click.Context,
+    title: str,
+    compare_out: str,
+    out_html: str,
+):
+    config = plot_compare.Config(
+        verbosity=2 if ctx.obj["verbose"] else 1,
+        compare_out=compare_out,
+        out_html=out_html,
+        title=title,
+    )
+    plot_compare.run(config)
+
+
+@cli.command("plot-var-het", help="Plot var(het) metric from .npz files.")
+@click.option(
+    "--title", default="NGS Chew var(het) Plot", help="title to use for the output HTML file."
+)
+@click.argument("stats_out")
+@click.argument("out_html")
+@click.pass_context
+def cli_plot_var_het(
+    ctx: click.Context,
+    title: str,
+    stats_out: str,
+    out_html: str,
+):
+    config = plot_var_het.Config(
+        verbosity=2 if ctx.obj["verbose"] else 1,
+        title=title,
+        out_html=out_html,
+        stats_out=stats_out,
+    )
+    plot_var_het.run(config)

chew/compare.py

@@ -1,13 +1,31 @@
 import math
+import typing
 
+import attrs
 from logzero import logger
 import numpy as np
 from tqdm import tqdm
 
 
-def load_fingerprint(path):
+@attrs.frozen
+class Config:
+    verbosity: int
+    output_prefix: str
+    min_mask_ones: typing.Optional[int]
+    max_mask_ones: typing.Optional[int]
+    fingerprints: typing.List[str]
+
+
+def load_fingerprint(path, *, load_aafs: bool = False):
     nparr = np.load(path)
-    return nparr["header"][3], nparr["fingerprint"]
+    if load_aafs:
+        return nparr["header"][3], nparr["fingerprint"], nparr.get("aafs")
+    else:
+        return nparr["header"][3], nparr["fingerprint"]
+
+
+def load_fingerprint_with_aafs(path):
+    return load_fingerprint(path, load_aafs=True)
 
 
 def relatedness(lhs, rhs):
@@ -35,18 +53,18 @@ def relatedness(lhs, rhs):
     return n_ibs0, rel, np.count_nonzero(mask)
 
 
-def run(args):
+def run(config: Config):
     logger.info("Loading fingerprints...")
     fps = {
-        name: fingerprint for name, fingerprint in map(load_fingerprint, tqdm(args.fingerprints))
+        name: fingerprint for name, fingerprint in map(load_fingerprint, tqdm(config.fingerprints))
     }
     logger.info("Loaded %d fingerprints", len(fps))
-    if args.min_mask_ones or args.max_mask_ones:
+    if config.min_mask_ones or config.max_mask_ones:
         fps = {
             name: fp
             for name, fp in fps.items()
-            if (not args.min_mask_ones or np.count_nonzero(fp[0]) >= args.min_mask_ones)
-            and (not args.max_mask_ones or np.count_nonzero(fp[0]) <= args.max_mask_ones)
+            if (not config.min_mask_ones or np.count_nonzero(fp[0]) >= config.min_mask_ones)
+            and (not config.max_mask_ones or np.count_nonzero(fp[0]) <= config.max_mask_ones)
         }
     logger.info("Filtered to %d fingerprints", len(fps))
     header = ["name"] + list(fps)
@@ -66,12 +84,12 @@ def run(args):
         rows_rel.append(row_rel)
         rows_mask.append(row_mask)
 
-    with open("%s.relatedness.txt" % args.output_prefix, "wt") as outputf:
+    with open("%s.relatedness.txt" % config.output_prefix, "wt") as outputf:
         for row in [header] + rows_rel:
             print("\t".join(map(str, row)), file=outputf)
-    with open("%s.ibs0.txt" % args.output_prefix, "wt") as outputf:
+    with open("%s.ibs0.txt" % config.output_prefix, "wt") as outputf:
         for row in [header] + rows_n_ibs0:
             print("\t".join(map(str, row)), file=outputf)
-    with open("%s.mask.txt" % args.output_prefix, "wt") as outputf:
+    with open("%s.mask.txt" % config.output_prefix, "wt") as outputf:
         for row in [header] + rows_mask:
             print("\t".join(map(str, row)), file=outputf)