model/include/silent_tools/silentappendwaters

#!/usr/bin/env python

import distutils.spawn
import os
import sys
sys.path.append(os.path.dirname(distutils.spawn.find_executable("silent_tools.py")))
import silent_tools
from silent_tools import eprint
from multiprocessing import Pool

# Don't throw an error when someone uses head
from signal import signal, SIGPIPE, SIG_DFL
signal(SIGPIPE, SIG_DFL) 

# This code assumes that the target backbone is fixed and will need major revisions if that is every changed - NRB
if (len(sys.argv) == 1):
    eprint("")
    eprint('silentappendwaters by Nate - a tool to allow you to add waters directly to a silent file')
    eprint("Usage:")
    eprint("        silentappendwaters solvated_centered_target.silent file_of_docks.silent > file_of_solvated_docks.silent")
    eprint("")
    eprint(" WARNING!!! This only works in the very specific context of adding waters to a miniprotein binder dock where" )
    eprint(" the waters are assigned chain B (the target's chain). If you find the need to use this is another case, talk" )
    eprint("                                     to Nate (nrbennet@uw.edu) ")
    sys.exit(1)

water_file = sys.argv[1]
silent_file = sys.argv[2]
n_cores = 1
if len(sys.argv) > 3:
    n_cores = int(sys.argv[3])
    try:
        available = int(silent_tools.cmd( "echo $SLURM_CPUS_ON_NODE" ))
    except():
        eprint("Parallel only available on a qlogin node")
        sys.exit( 0 )
    if n_cores > available:
        eprint( "More cpus requested (%i) than available (%i)"%(n_cores, available) )
        sys.exit( 0 )

silent_index = silent_tools.get_silent_index( silent_file )

( RT_lines, xyz_lines, ann_seq_append, seq_append, edge_list ) = silent_tools.get_water_info( water_file )

# I'm going to do this in 1000 struct chunks to minimize the amount of data being stored. this can easily be changed
seek_size = 1000
cur_struct = 0
ifile = 0
total_structures = len(silent_index['tags'])
header = silent_tools.silent_header( silent_index ).split('\n')
header[0] += seq_append

sys.stdout.write( '\n'.join(header) )
sys.stdout.flush()

parallel = n_cores > 1
if parallel:
    pool = Pool( n_cores )

with open( silent_file ) as open_silent_file:
    while cur_struct < total_structures:

        if ( cur_struct + seek_size > total_structures ):
            seek_size = total_structures - cur_struct

        structures = silent_tools.get_silent_structures_true_slice( open_silent_file, 
                            silent_index, cur_struct, cur_struct+seek_size, True )

        for structure in structures:
            if parallel:
                retval = pool.apply_async( silent_tools.solvate, args = [structure.split('\n'), RT_lines, xyz_lines, ann_seq_append, seq_append, edge_list] )
                solvated_structure = retval.get()
            else:
                solvated_structure = silent_tools.solvate( structure.split('\n'), RT_lines, xyz_lines, ann_seq_append, seq_append, edge_list )
            sys.stdout.write("\n".join(solvated_structure))
            sys.stdout.flush()

        cur_struct += seek_size

if parallel:
    pool.close()
    pool.join()