Merge pull request #13 from ncsa/develop

Adding SeedableRNG and improving error handling.

Cargo.toml

@@ -14,3 +14,5 @@ clap = { version = "4.5.1", features = ["derive"] }
 itertools = "0.12.1"
 assert_fs = "1.1.1"
 rand_distr = "0.5.0-alpha.0"
+rand_core = "0.9.0-alpha.0"
+rand_chacha = "0.9.0-alpha.0"

src/main.rs

@@ -5,6 +5,8 @@ extern crate simplelog;
 extern crate serde_yaml;
 extern crate rand_distr;
 extern crate itertools;
+extern crate rand_core;
+extern crate rand_chacha;
 
 mod utils;
 
@@ -13,14 +15,16 @@ use std::fs::File;
 use clap::{Parser};
 use log::*;
 use simplelog::*;
+use rand::SeedableRng;
 use rand::thread_rng;
-
+use rand_core::RngCore;
 use utils::cli;
 use utils::config::{read_config_yaml, build_config_from_args};
 use utils::file_tools::check_parent;
 use utils::runner::run_neat;
+use utils::neat_rng::NeatRng;
 
-fn main() -> Result<(), std::fmt::Error> {
+fn main() {
 
     info!("Begin processing");
     // parse the arguments from the command line
@@ -59,8 +63,6 @@ fn main() -> Result<(), std::fmt::Error> {
         )
     ]).unwrap();
 
-    let mut rng = thread_rng();
-
     // set up the config struct based on whether there was an input config. Input config
     // overrides any other inputs.
     let config = if args.config != "" {
@@ -69,10 +71,26 @@ fn main() -> Result<(), std::fmt::Error> {
     } else {
         info!("Using command line arguments.");
         debug!("Command line args: {:?}", &args);
-        Ok(build_config_from_args(args).expect("Problem reading configuration yaml file"))
-    }.unwrap();
+        build_config_from_args(args).unwrap_or_else(|error| {
+            panic!("Problem reading configuration yaml file {:?}", error)
+        })
+    };
+
+    // Generate the RNG used for this run. If one was given in the config file, use that, or else
+    // use thread_rng to generate a random seed, then seed using a SeedableRng based on StdRng
+    let seed: u64;
+    if !config.rng_seed.is_none() {
+        seed = config.rng_seed.unwrap();
+    } else {
+        // We pick a random u64 to use as a seed
+        let mut seed_rng = thread_rng();
+        seed = seed_rng.next_u64();
+    }
+    info!("Generating random numbers using the seed: {}", seed);
+    let mut rng: NeatRng = SeedableRng::seed_from_u64(seed);
 
-    run_neat(config, &mut rng).unwrap();
-    Ok(())
+    run_neat(config, &mut rng).unwrap_or_else(|error| {
+        panic!("Neat encountered a problem: {:?}", error)
+    })
 }
 

src/utils.rs

@@ -7,5 +7,5 @@ pub mod mutate;
 pub mod fastq_tools;
 pub mod vcf_tools;
 pub mod nucleotides;
-
+pub mod neat_rng;
 pub mod runner;

src/utils/cli.rs

@@ -1,9 +1,9 @@
-/// This is a pretty basic implementation of Clap CLI
-/// The idea of this interface and supporting code is that the user can enter an optional
-/// config file that will take the place of the other command line options, except the logging
-/// features, which are handled separately. Either way, these options are read into a configuration
-/// struct that holds the variables for the run. Logging, meanwhile, is handled separately,
-/// outside run configuration parsing.
+// This is a pretty basic implementation of Clap CLI
+// The idea of this interface and supporting code is that the user can enter an optional
+// config file that will take the place of the other command line options, except the logging
+// features, which are handled separately. Either way, these options are read into a configuration
+// struct that holds the variables for the run. Logging, meanwhile, is handled separately,
+// outside run configuration parsing.
 
 extern crate clap;