feat: implementing "seqvars prefilter" (#209)

README.md

@@ -18,6 +18,7 @@ At the moment, the following sub commands exist:
     - `db mk-inhouse` -- compile per-case structural variant into an in-house database previously created by `db compile`
 - `seqvars` -- subcommands for processing sequence (aka small/SNV/indel) variants
     - `seqvars ingest` -- convert single VCF file into internal format for use with `seqvars query`
+    - `seqvars prefilter` -- limit the result of `seqvars prefilter` by population frequency and/or distance to exon
     - `seqvars query` -- perform sequence variant filtration and on-the-fly annotation
 - `strucvars` -- subcommands for processing structural (aka large variants, CNVs, etc.) variants
     - `strucvars ingest` -- convert one or more structural variant files for use with `strucvars query`
@@ -129,6 +130,39 @@ Overall, the command will emit the following header rows in addition to the `##c
 > [!NOTE]
 > Future versions of the worker will annotate the worst effect on a MANE select or MANE Clinical transcript.
 
+## The `seqvars prefilter` Command
+
+This file takes as the input a file created by `seqvars ingest` and filters the variants by population frequency and/or distance to exon.
+You can pass the prefilter criteria as JSON on the command line corresponding to the following Rust structs:
+
+```rust
+struct PrefilterParams {
+    /// Path to output file.
+    pub path_out: String,
+    /// Maximal allele population frequency.
+    pub max_freq: f64,
+    /// Maximal distance to exon.
+    pub max_dist: i32,
+}
+```
+
+You can either specify the parameters on the command line directly or pass a path to a JSONL file starting with `@`.
+You can mix both ways.
+
+```
+$ varfish-server-worker strucvars prefilter \
+    --path-input INPUT.vcf \
+    --params '{"path_out": "out.vcf", "max_freq": 0.01, "max_dist": 100}' \
+    [--params ...] \
+
+# OR
+
+$ varfish-server-worker strucvars prefilter \
+    --path-input INPUT.vcf \
+    --params @path/to/params.json \
+    [--params ...] \
+```
+
 ## The `strucvars ingest` Command
 
 This command takes as the input one or more VCF files from structural variant callers and converts it into a file for further querying.

src/main.rs

@@ -84,6 +84,7 @@ struct Seqvars {
 #[derive(Debug, Subcommand)]
 enum SeqvarsCommands {
     Ingest(seqvars::ingest::Args),
+    Prefilter(seqvars::prefilter::Args),
 }
 
 fn main() -> Result<(), anyhow::Error> {
@@ -121,6 +122,9 @@ fn main() -> Result<(), anyhow::Error> {
                 SeqvarsCommands::Ingest(args) => {
                     seqvars::ingest::run(&cli.common, args)?;
                 }
+                SeqvarsCommands::Prefilter(args) => {
+                    seqvars::prefilter::run(&cli.common, args)?;
+                }
             },
             Commands::Strucvars(strucvars) => match &strucvars.command {
                 StrucvarsCommands::Ingest(args) => {

src/seqvars/mod.rs

@@ -1 +1,2 @@
 pub mod ingest;
+pub mod prefilter;

src/seqvars/prefilter/mod.rs

@@ -0,0 +1,31 @@
+//! Implementation of `seqvars prefilter` subcommand.
+
+use crate::common;
+
+/// Command line arguments for `seqvars prefilter` subcommand.
+#[derive(Debug, clap::Parser)]
+#[command(author, version, about = "prefilter an ingested variant VCF", long_about = None)]
+pub struct Args {
+    /// Path to input file.
+    #[clap(long)]
+    pub path_in: String,
+    /// Prefilter parameters or @ with path to JSONL file.
+    #[clap(long)]
+    pub params: Vec<String>,
+}
+
+/// Main entry point for `seqvars prefilter` sub command.
+pub fn run(args_common: &crate::common::Args, args: &Args) -> Result<(), anyhow::Error> {
+    let before_anything = std::time::Instant::now();
+    tracing::info!("args_common = {:#?}", &args_common);
+    tracing::info!("args = {:#?}", &args);
+
+    common::trace_rss_now();
+
+
+    tracing::info!(
+        "All of `seqvars ingest` completed in {:?}",
+        before_anything.elapsed()
+    );
+    Ok(())
+}