Skip to content

Instantly share code, notes, and snippets.

@samuell
Last active August 29, 2015 14:04
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 1 You must be signed in to fork a gist
  • Save samuell/6f0ecbe4c5e88c04c387 to your computer and use it in GitHub Desktop.
Save samuell/6f0ecbe4c5e88c04c387 to your computer and use it in GitHub Desktop.
use std::io::BufferedReader;
use std::io::File;
use std::str::StrSlice;
fn main() {
let path = Path::new("Homo_sapiens.GRCh37.67.dna_rm.chromosome.Y.fa");
let mut file = BufferedReader::new(File::open(&path));
let mut gc = 0i;
let mut at = 0i;
for line in file.lines() {
for c in line.unwrap().as_slice().chars() {
match c {
'G' => gc += 1,
'C' => gc += 1,
'A' => at += 1,
'T' => at += 1,
_ => {}
}
}
}
let gc_frac: f64 = (gc as f64) / ((at as f64) + (gc as f64));
println!("GC fraction: {}", gc_frac)
}
@dahankzter
Copy link

C still kix ass.... :D

@samuell
Copy link
Author

samuell commented Aug 6, 2014

C rocks ... as long as program safety is not that important :)

@Dr-Emann
Copy link

Dr-Emann commented Aug 7, 2014

I think the use of lines is slowing you down. It has to ensure the stream is valid utf8. It might be better to loop over .bytes() instead.

@huonw
Copy link

huonw commented Aug 7, 2014

The following is about 3× faster for me:

use std::io::File;

fn main() {
    let path = Path::new("Homo_sapiens.GRCh37.67.dna_rm.chromosome.Y.fa");
    let mut file = File::open(&path);
    let mut gc = 0i;
    let mut at = 0i;
    // large buffer of bytes
    let mut buf = [0u8, .. 1 << 16];
    loop {
        let n = match file.read(buf) {
            Ok(n) => n,

            // EOF etc.
            Err(_) => break
        };

        for b in buf.slice_to(n).iter() {
            match *b {
                b'G' => gc += 1,
                b'C' => gc += 1,
                b'A' => at += 1,
                b'T' => at += 1,
                _ => {}
            }
        }
    }

    let gc_frac: f64 = (gc as f64) / ((at as f64) + (gc as f64));
    println!("GC fraction: {}", gc_frac)
}

(Unlike Go, Rust ensures that strings are valid UTF8: the validator shows up high the profiles when using lines.)

@samuell
Copy link
Author

samuell commented Aug 8, 2014

@hounw: Cool, that's a 25% speedup of the fastest Go version I have on http://saml.rilspace.org/moar-languagez-gc-content-in-python-d-fpc-c-and-c ! :) Good job!

@samuell
Copy link
Author

samuell commented Aug 8, 2014

Hope to post updated benchmarks and graphs soon!

@samuell
Copy link
Author

samuell commented Aug 8, 2014

Aha, only I had forgot the check on whether each line starts with '>', which we have in https://gist.github.com/samuell/5591369 ... Do you think you could update your code with that, @hounw? (I'm unfortunately not versed in rust enough to know how to do it ...)

@huonw
Copy link

huonw commented Aug 8, 2014

@samuell

use std::io::File;

fn main() {
    let path = Path::new("Homo_sapiens.GRCh37.67.dna_rm.chromosome.Y.fa");
    let mut file = File::open(&path);
    let mut gc = 0i;
    let mut at = 0i;
    // large buffer of bytes
    let mut buf = [0u8, .. 1 << 16];

    let mut start_of_line = true;
    let mut inside_comment = false;

    loop {
        let n = match file.read(buf) {
            Ok(n) => n,

            // EOF etc.
            Err(_) => break
        };

        for b in buf.slice_to(n).iter() {
            let is_newline = *b == b'\n';
            if inside_comment {
                if is_newline {
                    inside_comment = false;
                }
            } else {
                match *b {
                    b'G' | b'C' => gc += 1,
                    b'A' | b'T' => at += 1,
                    b'>' if start_of_line => inside_comment = true,
                    _ => {}
                }
            }
            start_of_line = is_newline;
        }
    }

    let gc_frac: f64 = (gc as f64) / ((at as f64) + (gc as f64));
    println!("GC fraction: {}", gc_frac)
}

By the way, it seems some programs in your blog post skip the first line, and some others don't handle > line comments.

@pythonesque
Copy link

Did you catch the memory mapped version at https://gist.github.com/Blei/32d22fb92a3365da86b6?

@samuell
Copy link
Author

samuell commented Aug 6, 2015

@huonw: I think you're right about some unfortunate differences in the code examples. I really should compile an updated benchmark soon.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment