1.CodingFraction_dist.sh

#!/bin/bash


#####################################
#### Filter putative mtDNA contigs (works also on scaffolds) based on the coding fraction. 
#### Takes as input a directory containing all the files generated by the MITOS webserver.    
#### From the MITOS webserver, download the raw data folder. 
#### For each contig in each sample, there is a different raw data folder.
####
#### Usage: bash 1.CodingFraction Directory Prefix Output 
####        - Directory = directory with all MITOS raw data folders                               
####        - Prefix = ID of the sample being processed
####        - Output = Complete path to the output folder of your choice
####
#### Francesco Cicconardi (2016)
#####################################


### Getting script arguments
MITODIR=$1                
PREFIX=$2
OUTPUT=$3

### Changing directory, remove old tmp
cd $MITODIR
rm tmp.coding.frac


### Looping over alls contigs to calculate the coding fraction
ls -1 | while read contig
do
    cd $contig
    export SCF_len=`grep -v '^>' sequence.fas | awk '{sum+=(length$1)} END {print sum}'`
    export ANN_len=`grep gene result | cut -f 5,6 | awk '{sum+=($2-$1)} END {print sum+0.00000000001}'`
    echo $contig $ANN_len $SCF_len | awk '{frac=$2/$3} END {print $1, frac}' >> $OUTPUT/tmp.coding.frac
    cd ..
done

cd ..

### Generating output
cat $OUTPUT/tmp.coding.frac | sort -k2,2rg > $OUTPUT/$PREFIX.Coding_len_frac.dat

### Removing mess
rm $OUTPUT/tmp.coding.frac