Structural Change!

Changing script structure to have functions in separate files!

recipe/src/bye.sh

@@ -0,0 +1,13 @@
+read -r -d '' BYE << EOM
+
+ # Bye Bye
+ Now your plot is complete and must be available at: ${RESULTS}/conf/
+
+ All the required files for a minimal circos plot have been produced and stored at ${RESULTS}/conf/.
+ Now you can play with the ${RESULTS}/conf/circos.conf file in order to change the plot as you like.
+
+ Remember to read the circos manual in order to understand the conf file.
+
+ Have fun!
+
+EOM

recipe/src/check_links.sh

@@ -0,0 +1,13 @@
+check_links()
+{
+# get chrs with links
+## chr source
+cut -f 1 ${RESULTS}/conf/links_concatenated_colored.txt >> tmp.chrs
+## chr target
+cut -f 4 ${RESULTS}/conf/links_concatenated_colored.txt >> tmp.chrs
+
+# export LINE
+CHRS=$(cat tmp.chrs | sort -u | tr '\n' ';') ;
+rm tmp.chrs ;
+export CUSTOM_CHR_LINE="chromosomes = "${CHRS} ;
+}

recipe/src/dedup.sh

@@ -0,0 +1,8 @@
+dedup()
+{
+for file in ${RESULTS}/conf/*.txt; do
+    sort -u ${file} > tmp.txt ;
+    cat tmp.txt   > ${file} ;
+    rm tmp.txt
+done
+}

recipe/src/filter.sh

@@ -0,0 +1,17 @@
+filter()
+{
+# create results dir
+mkdir -p $RESULTS
+
+# create dir for files
+mkdir -p ${RESULTS}/filtered
+
+# filter genomes
+IFS=','
+while read -r FASTA FASTA_PREFIX FASTA_COLOR ; do
+  name="$(basename $FASTA)" ;
+  $CONDA_PREFIX/bin/perl $CONDA_PREFIX/bin/removesmalls.pl $MINLEN $FASTA >> ${RESULTS}/filtered/"$name" ;
+  continue
+done<"$FOFN"
+
+}

recipe/src/find_links.sh

@@ -0,0 +1,22 @@
+find_links()
+{
+# create dir
+mkdir -p ${RESULTS}/all_vs_all_blast
+
+# concatenate genomes
+cat ${RESULTS}/filtered/* >> ${RESULTS}/all_vs_all_blast/concatenated_genomes.fasta ;
+export CONCAT_FASTA=${RESULTS}/all_vs_all_blast/concatenated_genomes.fasta
+export BLAST_DB=${RESULTS}/all_vs_all_blast/blast_db
+
+# Run blast
+makeblastdb -in $CONCAT_FASTA -dbtype nucl -out $BLAST_DB &> /dev/null ;
+blastn -task blastn -perc_identity $MINID -query $CONCAT_FASTA -db $BLAST_DB \
+    -outfmt "6 qseqid qstart qend sseqid sstart send pident length mismatch gapopen evalue bitscore stitle" \
+    -out ${RESULTS}/all_vs_all_blast/tmp.blast -num_threads $THREADS
+
+# Filter blast
+awk -F '\t' -v minid=$MINID '{ if ($7 >= minid) { print } }' ${RESULTS}/all_vs_all_blast/tmp.blast > ${RESULTS}/all_vs_all_blast/all_vs_all.blast
+
+# Remove tmp
+rm ${RESULTS}/all_vs_all_blast/tmp.blast
+}

recipe/src/gc_skew.sh

@@ -0,0 +1,6 @@
+gc_skew()
+{
+# exec GCcalc.py
+$CONDA_PREFIX/bin/python3 $CONDA_PREFIX/bin/GCcalc.py -f ${RESULTS}/all_vs_all_blast/concatenated_genomes.fasta -w $GCWINDOW -s $GCSTEP | \
+    cut -f 1,2,3,5 | awk '{ if ($4 > 0) print $0 "\t" "color=dblue"; else print $0 "\t" "color=red"}' > ${RESULTS}/conf/GC_skew.txt
+}

recipe/src/help.sh

@@ -0,0 +1,53 @@
+Help()
+{
+cat << EOF
+
+Simple script to create a circos plot between two FASTA files.
+Copyright, Felipe Almeida <almeidafmarques@outlook.com>, 2021
+
+ Syntax: plot_circos.sh [-h] [--fofn <file> --outdir <outdir> --minlen <int> --minid <int>
+                              --linklen <int> --show_intrachr --gc_window <int> --gc_step <int>
+                              --labels <file> ]
+
+ Options:
+
+ # Help
+ -h/--help                           Print this help
+
+ # Threads for blastn
+ --threads                          Number of threads to use [Default: 1]
+
+ # Output
+ --outdir                            Path to output directory [Default: ./results]
+
+ # Input file of file names
+ # CSV: fasta path,prefix,color
+ --fofn                              File of file names contatining list of fastas to
+                                     draw circos plot.
+
+ # Input min. length
+ --minlen                            Min size of contigs to consider for plot [Default: 10000]
+
+ # Links (blastn) min. percentage id
+ --minid                             Min. percentage id to filter the results of blastn to draw links [Default: 85]
+ --linklen                           Min. link (blastn hit) length to display in plot [Default: 5000]
+ --show_intrachr                     Tells the program to create a conf file showing intra chr links [Default: false]
+                                     Mandatory if using only one FASTA, otherwise, links will not be shown.
+
+ # GC skew config
+ --gc_window                         GC skew window size [Default: 5000]
+ --gc_step                           GC skew step size [Default: 5000]
+
+ # Labels config
+ --labels                            TSV file containing the label definitions for plotting. The file must contain
+                                     3 or 4 columns as shown at http://circos.ca/documentation/tutorials/2d_tracks/text_1/lesson,
+                                     "DATA FORMAT" section. The first column must be the name (ID) of the contig.
+
+ # Tiles config
+ --tiles                             TSV file containing the tile definitions for plotting. The file must contain
+                                     3 or 4 columns as shown at http://circos.ca/documentation/tutorials/configuration/data_files.
+                                     The first column must be the name (ID) of the contig.
+
+
+EOF
+}

recipe/src/karyotypes.sh

@@ -0,0 +1,15 @@
+karyotype()
+{
+# create dir
+mkdir -p ${RESULTS}/conf
+
+# write karyotypes
+IFS=','
+while read -r FASTA FASTA_PREFIX FASTA_COLOR ; do
+  name="$(basename $FASTA)" ;
+  FILTERED_FASTA=${RESULTS}/filtered/"$name" ;
+  bioawk -c fastx -v p=$FASTA_PREFIX -v color=$FASTA_COLOR \
+    '{ printf "chr - " substr($name,1) " " p":" substr($name,1) " " "0" " " length($seq) " " color"\n" }' \
+    $FILTERED_FASTA >> ${RESULTS}/conf/circos.sequences.txt ;
+done<"$FOFN"
+}

recipe/src/labels.sh

@@ -0,0 +1,28 @@
+labels()
+{
+# Create label definition
+read -r -d '' LABELS_CONF << EOM
+# Labels
+# to understand it more read: http://circos.ca/documentation/tutorials/2d_tracks/text_1/lesson
+<plot>
+label_snuggle    = yes
+type             = text
+color            = black
+file             = ${LABELS}
+
+r0 = 1r
+r1 = 1r+200p
+
+show_links     = yes
+link_dims      = 4p,4p,8p,4p,4p
+link_thickness = 5p
+link_color     = black
+
+label_size   = 30p
+
+padding  = 0p
+rpadding = 0p
+
+</plot>
+EOM
+}

recipe/src/parse_links.sh

@@ -0,0 +1,21 @@
+parse_links()
+{
+# create dir
+mkdir -p ${RESULTS}/conf
+
+# Filter blocks with more then N bp hits
+awk -v minlen=$MINLINKLEN '{ if ($8 >= minlen) { print } }' \
+  ${RESULTS}/all_vs_all_blast/all_vs_all.blast | cut -f 1,2,3,4,5,6 >> ${RESULTS}/conf/links_concatenated.txt ;
+
+# get links comming from contigs and give it colors
+IFS=','
+while read -r FASTA FASTA_PREFIX FASTA_COLOR ; do
+  bioawk -c fastx '{ printf $name"\n" }' $FASTA > tmp_names.fasta ;
+  awk -v color1=$FASTA_COLOR -F'\t' 'NR==FNR{c[$1]++;next};c[$1] > 0 {print $0 "\t" "color="color1}' \
+    tmp_names.fasta ${RESULTS}/conf/links_concatenated.txt >> ${RESULTS}/conf/links_concatenated_colored.txt
+  rm tmp_names.fasta ;
+done<"$FOFN"
+
+# create additional file whithout intrachr links
+awk -F'\t' '{ if ($1 != $4) { print } }' ${RESULTS}/conf/links_concatenated_colored.txt > ${RESULTS}/conf/links_concatenated_colored_no_intrachr.txt ;
+}

recipe/src/plot_circos.sh

@@ -0,0 +1,14 @@
+plot_circos()
+{
+# get current dir
+CURRENT_DIR=$PWD
+
+# got to conf dir
+cd ${RESULTS}/conf/
+
+# draw
+circos
+
+# go back
+cd $CURRENT_DIR
+}

recipe/src/tiles.sh

@@ -0,0 +1,31 @@
+tiles()
+{
+# Create label definition
+read -r -d '' TILES_CONF << EOM
+# Tiles
+# to understand its configuration read: http://circos.ca/documentation/tutorials/2d_tracks/tiles/configuration
+<plot>
+type              = tile
+layers_overflow   = grow
+file              = ${TILES}
+r1                = 0.8r
+r0                = 0.7r
+orientation       = out
+
+layers      = 5
+margin      = 0.02u
+thickness   = 15
+padding     = 8
+
+stroke_thickness = 1
+stroke_color     = grey
+
+<backgrounds>
+<background>
+color = vvlgrey
+</background>
+</backgrounds>
+
+</plot>
+EOM
+}

recipe/src/workflow.sh

@@ -0,0 +1,58 @@
+workflow()
+{
+  ## remove existing results
+  rm -rf $RESULTS ;
+
+  # Step 1
+  echo " # Preparing inputs!"
+  filter          ;
+
+  # Step 2
+  echo " # Writing karyotypes!"
+  karyotype       ;
+
+  # Step 3
+  echo " # Finding links (all vs all blastn)!"
+  find_links      ;
+  parse_links     ;
+
+  # Step 4
+  echo " # Removing duplicate lines in conf files!"
+  dedup           ;
+  check_links     ;
+
+  # Step 5
+  echo " # Computing GC Skew!"
+  gc_skew         ;
+
+  # Step 6
+  # Check for labels
+  if [ -z "$LABELS" ]
+  then
+    echo "" > /dev/null ;
+  else
+    echo " # Parsing labels!" ;
+    labels                    ;
+  fi
+
+  # Step 7
+  # Check for tiles
+  if [ -z "$TILES" ]
+  then
+    echo "" > /dev/null ;
+  else
+    echo " # Parsing tiles!" ;
+    tiles                    ;
+  fi
+
+  # Step 6
+  echo " # Wrinting circos conf file!"
+  write_circos > ${RESULTS}/conf/circos.conf ;
+
+  # Step 7
+  echo " # Plotting circos!"
+  plot_circos     ;
+
+  # Bye
+  echo ${BYE}
+}