sand_filter_master(1)

NAME

sand_filter_master - filter sequences for alignment in parallel

SYNOPSIS

sand_filter_master [options] sequences.cfa candidates.cand

DESCRIPTION

sand_filter_master is the first step in the SAND assembler. It reads in a body of sequences, and uses a linear-time algorithm to produce a list of candidate sequences to be aligned in detail by sand_align_master(1).

This program uses the Work Queue system to distributed tasks among processors. After starting sand_filter_master, you must start a number of work_queue_worker(1) processes on remote machines. The workers will then connect back to the master process and begin executing tasks. The actual filtering is performed by sand_filter_kernel(1) on each machine.

OPTIONS

-p <port>
Port number for queue master to listen on. (default: 9123)
-s <size>
Number of sequences in each filtering task. (default: 1000)
-r <file>
A meryl file of repeat mers to be filtered out.
-R <n>
Automatically retry failed jobs up to n times. (default: 100)
-k <size>
The k-mer size to use in candidate selection (default is 22).
-w <size>
The minimizer window size. (default is 22).
-u
If set, do not unlink temporary binary output files.
-c <file>
Checkpoint filename; will be created if necessary.
-d <flag>
Enable debugging for this subsystem. (Try -d all to start.)
-F <number>
Work Queue fast abort multiplier. (default is 10.)
-Z <file>
Select port at random and write it out to this file.
-o <file>
Send debugging to this file.
-v
Show version string
-h
Show this help screen

EXIT STATUS

On success, returns zero. On failure, returns non-zero.

EXAMPLES

If you begin with a FASTA formatted file of reads, used sand_compress_reads(1) to produce a compressed FASTA (cfa) file. To run filtering sequentially, start a single work_queue_worker(1) process in the background. Then, invoke sand_filter_master.

% sand_compress_reads mydata.fasta mydata.cfa
% work_queue_worker localhost 9123 &
% sand_filter_master mydata.cfa mydata.cand

To speed up the process, run more work_queue_worker(1) processes on other machines, or use condor_submit_workers(1) or sge_submit_workers(1) to start hundreds of workers in your local batch system.