hole-process_copy

#!/bin/bash

# Note: 21 Nov 2016
#1. Need to make code for 'create_inp'. Make sure pathways are correct. (Done! 22-11-16)
#2. Need to make code for 'G_stats.py'. Print out averages and std deviations.
#	Select most relevant quantities we want to keep track of.
#3. Test code to death! 

#Note: 22 Nov 2016
#1. Need to rename frame files to feature time-points[ps]
#2. Re-think of how many essays we need to get the std deviation on HOLE calculations
# Keep in in mind that the more essays, the longer the analysis will take.  

function create_inp {
# This function creates ai simple HOLE input file
fname=$1
echo "!HOLE input file
coord ${fname}.pdb
radius ../hole2/rad/simple.rad
ignore hoh
capsule
sphpdb ${fname}.sph
endrad 15." > ${fname}.inp
}

function avg_std {
	X=$@ # Data array
	a=$(for x in $X ; do echo $x; done | awk '{sum1+=$1} {sum2+=$1^2} END{print sum1/NR, (sum2/NR - sum1^2/NR^2)^0.5}')
	echo $a
}

function G_stats {
#This function determines the average and std dev. of conductance values and dimensions
#over a all HOLE trials for a particular frame 
frame=$1 # PDB file name

# Extract data
# Channel length [Angstroms]
Length=$(grep '"Atomic" length of channel:' ${frame%.pdb}_1trial.txt | awk '{print $5}')
# Channel Minimum Raiuds [Angstroms] 
Rmin=$(grep TAG ${frame%.pdb}_*trial.txt | awk '{print $5}')
# Non-corrected Integrated Conductance (Bulk) [pS]
Gmacro=$(grep TAG ${frame%.pdb}_*trial.txt | awk '{print $7}')
# Corrected Integrated Conductance by Rmin (8 system)  [pS]
Gpred_Rmin=$(grep TAG ${frame%.pdb}_*trial.txt | awk '{print $10 }')
# Corrected Integrated Conductance by Lenght  [pS] 
Gpred_Lenght=$(grep TAG ${frame%.pdb}_*trial.txt | awk '{print $11 }')
# Corrected Integrated Conductance by Average Electric Potential  [pS]
Gpred_AvEPot=$(grep TAG ${frame%.pdb}_*trial.txt | awk '{print $12}')

# Compute average and standard deviation values
x2=$(avg_std $Rmin)
x3=$(avg_std $Gmacro)
x4=$(avg_std $Gpred_Rmin)
x5=$(avg_std $Gpred_Lenght)
x6=$(avg_std $Gpred_AvEPot)
printf "%s \t %3f \t %s \t %s \t %s \t %s \t %s\n" $Length 0 "$x2" "$x3" "$x4" "$x5" "$x6"
} 

fname=$1 # Generic file name (without extension)
dt=$2    # Time-space between frames [ps]
mkdir ${fname}_frames
# Split trajectory into seprate frame fles(PDB format)
# Select only Protein coordinates 
echo "Protein" | gmx trjconv -f ${fname}.xtc -s ${fname}.tpr -dt $dt -sep -o ${fname}_frames/frame.pdb 
cd ${fname}_frames
for ffile in `ls frame*pdb`; do 
	# Create a HOLE input file per frame file
	create_inp ${ffile%.pdb}
	for n in `seq 1 1`; do  
		# Run HOLE for each frame file (5 independent trials)
		../hole2/exe/hole < ${ffile%.pdb}.inp > ${ffile%.pdb}_${n}trial.txt
	done
	output=$(G_stats ${ffile})
	echo "$output" >> ../HOLE-Summary.dat
	rm ${ffile%.pdb}*	
done
cd ../
rm -r ${fname}_frames