0.6.1 (#69)

Fix: the cli version string.

---------

Co-authored-by: Mitchell R. Vollger <[email protected]>

Cargo.toml

@@ -10,7 +10,7 @@ license = "MIT"
 name = "fibertools-rs"
 readme = "README.md"
 repository = "https://github.com/fiberseq/fibertools-rs"
-version = "0.6.0"
+version = "0.6.1"
 # exclude py-ft and test data from cargo publish since they are too large
 exclude = ["py-ft/", "tests/data/"]
 
@@ -57,7 +57,7 @@ rand = "0.8.5"
 
 [build-dependencies]
 burn-import = {version = "0.12"}
-vergen-git2 = { version = "1.0.1", features = ["build"] }
+vergen-git2 = { version = "1.0.2", features = ["build", "cargo", "rustc", "si"] }
 
 
 [dev-dependencies]

build.rs

@@ -6,11 +6,22 @@ fn vergen() -> Result<(), Box<dyn Error>> {
     // NOTE: See the specific builder documentation for configuration options.
     let build = BuildBuilder::all_build()?;
     let git2 = Git2Builder::all_git()?;
+    let cargo = CargoBuilder::all_cargo()?;
 
-    Emitter::default()
+    let status = Emitter::default()
         .add_instructions(&build)?
         .add_instructions(&git2)?
-        .emit()?;
+        .add_instructions(&cargo)?
+        .fail_on_error()
+        .emit();
+
+    // set the env variable myself if the status failed
+    if status.is_err() {
+        eprintln!("Failed to get version information from git");
+        eprintln!("Likely building a published version from cargo or bioconda.");
+        std::env::set_var("VERGEN_GIT_DESCRIBE", "unknown");
+    }
+
     Ok(())
 }
 

src/fiber.rs

@@ -62,7 +62,9 @@ impl FiberseqData {
         };
 
         // get fiberseq basemods
-        let base_mods = BaseMods::new(&record, filters.min_ml_score);
+        let mut base_mods = BaseMods::new(&record, filters.min_ml_score);
+        base_mods.filter_at_read_ends(filters.strip_starting_basemods);
+
         //let (m6a, cpg) = FiberMods::new(&base_mods);
         let m6a = base_mods.m6a();
         let cpg = base_mods.cpg();

src/lib.rs

@@ -20,7 +20,11 @@ use std::io::Write;
 
 pub const VERSION: &str = env!("CARGO_PKG_VERSION");
 lazy_static! {
-    pub static ref FULL_VERSION: String = format!("{}", env!("VERGEN_GIT_DESCRIBE"));
+    pub static ref FULL_VERSION: String = format!(
+        "v{}\tgit-details {}",
+        env!("CARGO_PKG_VERSION"),
+        env!("VERGEN_GIT_DESCRIBE")
+    );
 }
 // if this string (bar)gets too long it displays weird when writing to stdout
 const PROGRESS_STYLE: &str =

src/utils/bamranges.rs

@@ -162,6 +162,44 @@ impl Ranges {
         self.reference_lengths = to_keep.iter().map(|&i| self.reference_lengths[i]).collect();
     }
 
+    /// filter out ranges that are within the first or last X bp of the read
+    pub fn filter_starts_at_read_ends(&mut self, strip: i64) {
+        if strip == 0 {
+            return;
+        }
+        let to_keep = self
+            .starts
+            .iter()
+            .enumerate()
+            .filter_map(|(i, &s)| {
+                if let Some(s) = s {
+                    if s < strip || s > self.seq_len - strip {
+                        None
+                    } else {
+                        Some(i)
+                    }
+                } else {
+                    None
+                }
+            })
+            .collect::<Vec<_>>();
+
+        if to_keep.len() != self.starts.len() {
+            log::trace!(
+                "basemods stripped, {} basemods removed",
+                self.starts.len() - to_keep.len()
+            );
+        }
+
+        self.starts = to_keep.iter().map(|&i| self.starts[i]).collect();
+        self.ends = to_keep.iter().map(|&i| self.ends[i]).collect();
+        self.lengths = to_keep.iter().map(|&i| self.lengths[i]).collect();
+        self.qual = to_keep.iter().map(|&i| self.qual[i]).collect();
+        self.reference_starts = to_keep.iter().map(|&i| self.reference_starts[i]).collect();
+        self.reference_ends = to_keep.iter().map(|&i| self.reference_ends[i]).collect();
+        self.reference_lengths = to_keep.iter().map(|&i| self.reference_lengths[i]).collect();
+    }
+
     pub fn to_strings(&self, reference: bool, skip_none: bool) -> Vec<String> {
         let (s, e, l, q) = if reference {
             (

src/utils/basemods.rs

@@ -51,6 +51,13 @@ impl BaseMod {
     pub fn is_cpg(&self) -> bool {
         self.modification_type == 'm'
     }
+
+    pub fn filter_at_read_ends(&mut self, n_strip: i64) {
+        if n_strip <= 0 {
+            return;
+        }
+        self.ranges.filter_starts_at_read_ends(n_strip);
+    }
 }
 
 #[derive(Eq, PartialEq, Debug, Clone)]
@@ -257,6 +264,16 @@ impl BaseMods {
             .for_each(|bm| bm.ranges.filter_by_qual(min_ml_score));
     }
 
+    /// filter the basemods at the read ends
+    pub fn filter_at_read_ends(&mut self, n_strip: i64) {
+        if n_strip <= 0 {
+            return;
+        }
+        self.base_mods
+            .iter_mut()
+            .for_each(|bm| bm.filter_at_read_ends(n_strip));
+    }
+
     /// combine the forward and reverse m6a data
     pub fn m6a(&self) -> Ranges {
         let ranges = self

src/utils/input_bam.rs

@@ -40,6 +40,15 @@ pub struct FiberFilters {
     /// Minium score in the ML tag to use or include in the output
     #[clap(long="ml", alias="min-ml-score", default_value = MIN_ML_SCORE, help_heading = "BAM-Options", env="FT_MIN_ML_SCORE")]
     pub min_ml_score: u8,
+    /// strip basemods in the first or last X bp of the read
+    #[clap(
+        global = true,
+        long,
+        default_value = "0",
+        help_heading = "BAM-Options",
+        hide = true
+    )]
+    pub strip_starting_basemods: i64,
 }
 
 impl std::default::Default for FiberFilters {
@@ -48,6 +57,7 @@ impl std::default::Default for FiberFilters {
             bit_flag: 0,
             min_ml_score: MIN_ML_SCORE.parse().unwrap(),
             filter_expression: None,
+            strip_starting_basemods: 0,
         }
     }
 }