Add bam support for counting intervals

gtars/Cargo.toml

@@ -20,12 +20,15 @@ ndarray-npy = "0.8.1"
 ndarray = "0.15.6"
 tempfile = "3.10.1"
 byteorder = "1.5.0"
-noodles = { version = "0.83.0", features = ["bam"] }
+noodles = { version = "0.83.0", features = ["bam", "sam", "bgzf"] }
 bstr = "1.10.0"
 rayon = "1.10.0"
 indicatif = "0.17.8"
+bigtools = "0.5.4"
+tokio = "1.40.0"
+os_pipe = "1.2.1"
 glob = "0.3.1"
-bigtools = "0.5.2"
+
 
 
 [dev-dependencies]

gtars/src/uniwig/README.md

@@ -2,10 +2,10 @@
 
 ### Input Bed File
 
-Currently, Uniwig accepts a single `.bed` file. It should be sorted by chromosome. This single bed file will be used to create 3 wiggle files (`.wig`):
-`_start.wig` -> accumulations of start coordinates
-`_end.wig` -> accumulations of end coordinates
-`_core.wig` -> accumulations of peaks (between starts and ends)
+Currently, Uniwig accepts a single `.bed` file, `.narrowPeak` file, `.bam` file. It should be sorted by chromosome. This single file will be used to create 3 output files:
+`_start` -> accumulations of start coordinates
+`_end` -> accumulations of end coordinates
+`_core` -> accumulations of peaks (between starts and ends)
 
 The below script can be used to create a sorted bed file from a directory of bed files:
 
@@ -39,30 +39,41 @@ The chrom.sizes reference is an optional argument. Uniwig will default to using
 
 ### Usage
 ```
-Create wiggle files from a BED or BAM file
+Create accumulation files from a BED or BAM file
 
 Usage: gtars uniwig [OPTIONS] --file <file> --smoothsize <smoothsize> --stepsize <stepsize> --fileheader <fileheader> --outputtype <outputtype>
 
 Options:
   -f, --file <file>              Path to the combined bed file we want to transform or a sorted bam file
-  -t, --filetype <filetype>      'bed' or 'bam' [default: bed]
+  -t, --filetype <filetype>      Input file type, 'bed' 'bam' or 'narrowpeak' [default: bed]
   -c, --chromref <chromref>      Path to chromreference
   -m, --smoothsize <smoothsize>  Integer value for smoothing
   -s, --stepsize <stepsize>      Integer value for stepsize
   -l, --fileheader <fileheader>  Name of the file
   -y, --outputtype <outputtype>  Output as wiggle or npy
-  -h, --help 
+  -u, --counttype <counttype>    Select to only output start, end, or core. Defaults to all. [default: all]
+  -p, --threads <threads>        Number of rayon threads to use for parallel processing [default: 6]
+  -o, --score                    Count via score (narrowPeak only!)
+  -z, --zoom <zoom>              Number of zoom levels (for bw file output only [default: 5]
+  -d, --debug                    Print more verbose debug messages?
+  -h, --help                     Print help
 
-```
 
-### Create bigwig files from wiggle files
+```
 
-Once you have created wiggle files, you can convert them to bigWig files using `wigToBigWig` (see: https://genome.ucsc.edu/goldenPath/help/bigWig.html, https://github.com/ucscGenomeBrowser/kent/tree/master/src/utils/wigToBigWig):
+### Processing bam files to bw
 
+Example command
 ```
-./wigToBigWig ./test_rust_wig/_end.wig ./sourcefiles/hg38.chrom.sizes ./end_rust.bw
+gtars uniwig -f "test1_chr1_chr2.bam" -m 5 -s 1 -l /myoutput/directory/test_file_name -y bw -t bam -p 6 -c /genome/alias/hg38/fasta/default/hg38.chrom.sizes -u all
+
 ```
 
+
 ### Export types
 
-Currently only `.wig` and `.npy` are supported as output types. 
+For Input types: `.bed` and `.narrowPeak`
+Output types include `.wig`, `.npy`, `.bedGraph`, and `.bw`
+
+For Input Types: `.bam`
+Output types include `.bw` and `.bed`

gtars/src/uniwig/cli.rs

@@ -9,7 +9,7 @@ use crate::uniwig::consts::UNIWIG_CMD;
 pub fn create_uniwig_cli() -> Command {
     Command::new(UNIWIG_CMD)
         .author("DRC")
-        .about("Create wiggle files from a BED or BAM file")
+        .about("Create accumulation files from a BED or BAM file")
         .arg(
             Arg::new("file")
                 .long("file")
@@ -61,6 +61,14 @@ pub fn create_uniwig_cli() -> Command {
                 .help("Output as wiggle or npy")
                 .required(true),
         )
+        .arg(
+            Arg::new("counttype")
+                .long("counttype")
+                .short('u')
+                .default_value("all")
+                .help("Select to only output start, end, or core. Defaults to all.")
+                .required(false),
+        )
         .arg(
             Arg::new("threads")
                 .long("threads")
@@ -81,9 +89,16 @@ pub fn create_uniwig_cli() -> Command {
             Arg::new("zoom")
                 .long("zoom")
                 .short('z')
-                .default_value("0")
+                .default_value("5")
                 .value_parser(clap::value_parser!(i32))
                 .help("Number of zoom levels (for bw file output only")
                 .required(false),
         )
+        .arg(
+            Arg::new("debug")
+                .long("debug")
+                .short('d')
+                .help("Print more verbose debug messages?")
+                .action(ArgAction::SetTrue),
+        )
 }