#6 Adding advanced options functionality to each config file and wrapper

ANGSD-wrapper · Mar 6, 2020 · c664bc7 · c664bc7
1 parent 6e626f9
commit c664bc7
Show file tree

Hide file tree

Showing 17 changed files with 204 additions and 81 deletions.
diff --git a/Configuration_Files/Abbababa_Config b/Configuration_Files/Abbababa_Config
@@ -54,3 +54,8 @@ DO_ABBABABA=1
 REMOVE_TRANS=0
 #       Set the size for each block
 BLOCKSIZE=1000
+
+#   For advanced users who want to change arguments used by ANGSD (i.e. which SAF method is used)
+#       Expected format is the same as how the flags and arguments are written on the command line:
+#       '-flag1 arg1 -flag2 arg2 ...' 
+ADVANCED_ARGS=''
diff --git a/Configuration_Files/Ancestral_Sequence_Config b/Configuration_Files/Ancestral_Sequence_Config
@@ -30,3 +30,8 @@ DO_FASTA=1
 #       If DO_FASTA is 2, DO_COUNTS must be 1
 #       Otherwise, DO_COUNTS can be any other legal value
 DO_COUNTS=0
+
+#   For advanced users who want to change arguments used by ANGSD (i.e. which SAF method is used)
+#       Expected format is the same as how the flags and arguments are written on the command line:
+#       '-flag1 arg1 -flag2 arg2 ...' 
+ADVANCED_ARGS=''
diff --git a/Configuration_Files/FST_Config b/Configuration_Files/FST_Config
@@ -85,3 +85,8 @@ GLOBAL=true
 WIN=1000
 #       Set the step size for sliding window analysis
 STEP=500
+
+#   For advanced users who want to change arguments used by ANGSD (i.e. which SAF method is used)
+#       Expected format is the same as how the flags and arguments are written on the command line:
+#       '-flag1 arg1 -flag2 arg2 ...' 
+ADVANCED_ARGS=''
diff --git a/Configuration_Files/Genotype_Config b/Configuration_Files/Genotype_Config
@@ -64,3 +64,8 @@ POST_CUTOFF=0.95
 SNP_PVAL=1e-6
 #       Output genotype likelihood frequency file
 DO_GLF=2
+
+#   For advanced users who want to change arguments used by ANGSD (i.e. which SAF method is used)
+#       Expected format is the same as how the flags and arguments are written on the command line:
+#       '-flag1 arg1 -flag2 arg2 ...' 
+ADVANCED_ARGS=''
diff --git a/Configuration_Files/Inbreeding_Coefficients_Config b/Configuration_Files/Inbreeding_Coefficients_Config
@@ -64,3 +64,8 @@ DO_GLF=3
 #       Set a seed value for creating approximate inbreeding coefficients
 #       Use the random number generator built into BASH
 SEED=$RANDOM
+
+#   For advanced users who want to change arguments used by ANGSD (i.e. which SAF method is used)
+#       Expected format is the same as how the flags and arguments are written on the command line:
+#       '-flag1 arg1 -flag2 arg2 ...' 
+ADVANCED_ARGS=''
diff --git a/Configuration_Files/Principal_Component_Analysis_Config b/Configuration_Files/Principal_Component_Analysis_Config
@@ -50,3 +50,8 @@ NORM=0
 CALL=0
 #       Set the maximum number of sites to use for Principal Component Analysis
 N_SITES=100000
+
+#   For advanced users who want to change arguments used by ANGSD (i.e. which SAF method is used)
+#       Expected format is the same as how the flags and arguments are written on the command line:
+#       '-flag1 arg1 -flag2 arg2 ...' 
+ADVANCED_ARGS=''
diff --git a/Configuration_Files/Site_Frequency_Spectrum_Config b/Configuration_Files/Site_Frequency_Spectrum_Config
@@ -66,3 +66,8 @@ DO_POST=1
 DO_SAF=2
 #       Overwrite any previously generated results
 OVERRIDE=true
+
+#   For advanced users who want to change arguments used by ANGSD (i.e. which SAF method is used)
+#       Expected format is the same as how the flags and arguments are written on the command line:
+#       '-flag1 arg1 -flag2 arg2 ...' 
+ADVANCED_ARGS=''
diff --git a/Configuration_Files/Thetas_Config b/Configuration_Files/Thetas_Config
@@ -73,3 +73,8 @@ SLIDING_WINDOW=false
 WIN=1000
 #       Set the step size for sliding window analysis
 STEP=500
+
+#   For advanced users who want to change arguments used by ANGSD (i.e. which SAF method is used)
+#       Expected format is the same as how the flags and arguments are written on the command line:
+#       '-flag1 arg1 -flag2 arg2 ...' 
+ADVANCED_ARGS=''
diff --git a/Wrappers/Abbababa.sh b/Wrappers/Abbababa.sh
@@ -47,8 +47,7 @@ then
     Rscript "${SOURCE}"/Wrappers/sortRegions.R "${REGIONS}" "${FAI}"
     REGIONS="$(find $(dirname ${REGIONS}) -name "*_sorted.txt")"
     echo "Running Abbababa" >&2
-    "${ANGSD_DIR}"/angsd \
-        -doAbbababa "${DO_ABBABABA}" \
+    WRAPPER_ARGS=$(echo -doAbbababa "${DO_ABBABABA}" \
         -rmTrans "${REMOVE_TRANS}" \
         -blockSize "${BLOCKSIZE}" \
         -doCounts "${DO_COUNTS}" \
@@ -60,13 +59,12 @@ then
         -minInd "${MIN_IND}" \
         -nThreads "${N_CORES}" \
         -rf "${REGIONS}" \
-        -out "${OUT}"/"${PROJECT}".D
+        -out "${OUT}"/"${PROJECT}".D)
 #   Are we missing a definiton for regions?
 elif [[ -z "${REGIONS}" ]]
 then
     echo "Running Abbababa" >&2
-    "${ANGSD_DIR}"/angsd \
-        -doAbbababa "${DO_ABBABABA}" \
+    WRAPPER_ARGS=$(echo -doAbbababa "${DO_ABBABABA}" \
         -rmTrans "${REMOVE_TRANS}" \
         -blockSize "${BLOCKSIZE}" \
         -doCounts "${DO_COUNTS}" \
@@ -77,12 +75,11 @@ then
         -minQ "${MIN_BASEQUAL}" \
         -minInd "${MIN_IND}" \
         -nThreads "${N_CORES}" \
-        -out "${OUT}"/"${PROJECT}".D
+        -out "${OUT}"/"${PROJECT}".D)
 #   Assuming a single reigon was defined in config file
 else
     echo "Running Abbababa" >&2
-    "${ANGSD_DIR}"/angsd \
-        -doAbbababa "${DO_ABBABABA}" \
+    WRAPPER_ARGS=$(echo -doAbbababa "${DO_ABBABABA}" \
         -rmTrans "${REMOVE_TRANS}" \
         -blockSize "${BLOCKSIZE}" \
         -doCounts "${DO_COUNTS}" \
@@ -94,8 +91,12 @@ else
         -minInd "${MIN_IND}" \
         -nThreads "${N_CORES}" \
         -r "${REGIONS}" \
-        -out "${OUT}"/"${PROJECT}".D
+        -out "${OUT}"/"${PROJECT}".D)
 fi
+# Check for advanced arguments, and overwrite any overlapping definitions
+FINAL_ARGS=$(source ${SOURCE}/Wrappers/Arg_Zipper.sh "${WRAPPER_ARGS}" "${ADVANCED_ARGS}")
+# echo "Final arguments: ${FINAL_ARGS}" 1<&2
+"${ANGSD_DIR}"/angsd ${FINAL_ARGS}
 
 #   jackKnife.R is provided with angsd.
 echo "Using jackKnife.R to finish Abbababa" >&2

diff --git a/Wrappers/Ancestral_Sequence.sh b/Wrappers/Ancestral_Sequence.sh
@@ -16,17 +16,19 @@ ANGSD_DIR=${SOURCE}/dependencies/angsd
 #       Check to see if we're using -doCounts
 if [[ -z "${DO_COUNTS}" ]]
 then
-    "${ANGSD_DIR}"/angsd \
-        -doFasta "${DO_FASTA}" \
+    WRAPPER_ARGS=$(echo -doFasta "${DO_FASTA}" \
         -i "${ANC_BAM}"\
         -out "${OUT}"
 else
-    "${ANGSD_DIR}"/angsd \
-        -doFasta "${DO_FASTA}" \
+    WRAPPER_ARGS=$(echo -doFasta "${DO_FASTA}" \
         -doCounts "${DO_COUNTS}" \
         -i "${ANC_BAM}"\
         -out "${OUT}"
 fi
+# Check for advanced arguments, and overwrite any overlapping definitions
+FINAL_ARGS=$(source ${SOURCE}/Wrappers/Arg_Zipper.sh "${WRAPPER_ARGS}" "${ADVANCED_ARGS}")
+# echo "Final arguments: ${FINAL_ARGS}" 1<&2
+"${ANGSD_DIR}"/angsd ${FINAL_ARGS}
 
 #   If we have SAMTools, might as well index
 if `command -v samtools > /dev/null 2> /dev/null`

diff --git a/Wrappers/Arg_Zipper.sh b/Wrappers/Arg_Zipper.sh
@@ -0,0 +1,68 @@
+#!/bin/bash
+
+set -e
+set -u
+
+
+echo -e "Zipping advanced arguments onto basic ones...\n" 1>&2
+wrapper_args=$1 # Which arguments are set by default?
+adv_args=$2 # Which arguments were defined by an advanced user?
+# Debugging outputs
+#     echo -e "Wrapper's: ${wrapper_args}\n" 1>&2
+#     echo -e "User's: ${adv_args}\n" 1>&2
+
+# Initialize lists for flags and values, and the final aggregate string
+FINAL_FLAGS=()
+FINAL_VALS=()
+FINAL_ARGS=""
+
+# Need to check existence of first flag-argument pair before entering while loop (~do-while loop)
+k=1
+flag=$(echo "${wrapper_args}" | cut -d " " -f $(( k )))
+val=$(echo "${wrapper_args}" | cut -d " " -f $(( k+1 )))
+while [[ ! -z $flag  && ! -z $val ]]
+do
+    FINAL_FLAGS+=("${flag}")
+    FINAL_VALS+=("${val}")
+    let k=k+2
+    flag=$(echo "${wrapper_args}" | cut -d " " -f $(( k )))
+    val=$(echo "${wrapper_args}" | cut -d " " -f $(( k+1 )))
+done
+
+# Need to check existence of first flag-argument pair before entering while loop (~do-while loop)
+k=1
+flag=$(echo "${adv_args}" | cut -d " " -f $(( k )))
+val=$(echo "${adv_args}" | cut -d " " -f $(( k+1 )))
+while [[ ! -z "${flag}"  && ! -z "${val}" ]]
+do
+    # Check for pre-existing flags in the wrapper's args that share the same name
+    # If such a flag exists, overwrite the associated value.
+    # Otherwise append the pair to the list
+    found=false
+    for i in "${!FINAL_FLAGS[@]}"
+    do
+	if [[ "${FINAL_FLAGS[i]}" = "${flag}" ]]
+	then
+	    echo "Found an overlapping flag '${flag}', overwriting value '${FINAL_VALS[i]}' with '${val}'" 1>&2
+	    FINAL_VALS[i]="${val}"
+	    found=true
+	fi
+    done
+
+    if [ $found = false ]
+    then
+	FINAL_FLAGS+=("${flag}")
+	FINAL_VALS+=("${val}")
+    fi
+    (( k=k+2 ))
+    flag=$(echo "${adv_args}" | cut -d " " -f $(( k )))
+    val=$(echo "${adv_args}" | cut -d " " -f $(( k+1 )))
+done
+
+for i in "${!FINAL_FLAGS[@]}"
+do
+    # echo -e "FINAL_FLAGS[${i}]: ${FINAL_FLAGS[i]}\t" 1<&2
+    FINAL_ARGS+="${FINAL_FLAGS[i]} ${FINAL_VALS[i]} "
+done
+
+echo "${FINAL_ARGS}"
diff --git a/Wrappers/FST.sh b/Wrappers/FST.sh
@@ -57,8 +57,7 @@ else
     if [[ -f "${REGIONS}" ]]
     then
         echo "WRAPPER: $GROUP_1 sfs starting..." >&2
-        "${ANGSD_DIR}"/angsd \
-            -bam "${G1_SAMPLE_LIST}" \
+	WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
             -out "${OUT}"/"${GROUP_1}"_Intergenic \
             -doMajorMinor "${DO_MAJORMINOR}" \
             -doMaf "${DO_MAF}" \
@@ -75,12 +74,11 @@ else
             -GL "${GT_LIKELIHOOD}" \
             -P "${N_CORES}" \
             -rf "${REGIONS}" \
-            -doPost "${DO_POST}"
+            -doPost "${DO_POST}")
     elif [[ -z "${REGIONS}" ]]
     then
         echo "WRAPPER: $GROUP_1 sfs starting" >&2
-        "${ANGSD_DIR}"/angsd \
-            -bam "${G1_SAMPLE_LIST}" \
+	WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
             -out "${OUT}"/"${GROUP_1}"_Intergenic \
             -doMajorMinor "${DO_MAJORMINOR}" \
             -doMaf "${DO_MAF}" \
@@ -96,11 +94,10 @@ else
             -ref "${REF_SEQ}" \
             -GL "${GT_LIKELIHOOD}" \
             -P "${N_CORES}" \
-            -doPost "${DO_POST}"
+            -doPost "${DO_POST}")
     else
         echo "WRAPPER: $GROUP_1 sfs starting" >&2
-        "${ANGSD_DIR}"/angsd \
-            -bam "${G1_SAMPLE_LIST}" \
+	WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
             -out "${OUT}"/"${GROUP_1}"_Intergenic \
             -doMajorMinor "${DO_MAJORMINOR}" \
             -doMaf "${DO_MAF}" \
@@ -116,9 +113,14 @@ else
             -ref "${REF_SEQ}" \
             -GL "${GT_LIKELIHOOD}" \
             -P "${N_CORES}" \
-            -r "${REGIONS}"
+            -r "${REGIONS}")
     fi
 fi
+# Check for advanced arguments, and overwrite any overlapping definitions
+FINAL_ARGS=$(source ${SOURCE}/Wrappers/Arg_Zipper.sh "${WRAPPER_ARGS}" "${ADVANCED_ARGS}")
+# echo "Final arguments: ${FINAL_ARGS}" 1<&2
+"${ANGSD_DIR}"/angsd ${FINAL_ARGS}
+
 
 #   For 2nd group:
 if [[ -f "${OUT}"/"${GROUP_2}_Intergenic.saf" ]] && [ "$OVERRIDE" = "false" ]
@@ -129,8 +131,7 @@ else
     if [[ -f "${REGIONS}" ]]
     then
         echo "WRAPPER: $GROUP_2 sfs starting..." >&2
-        "${ANGSD_DIR}"/angsd \
-            -bam "${G2_SAMPLE_LIST}" \
+	WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
             -out "${OUT}"/"${GROUP_2}"_Intergenic \
             -doMajorMinor "${DO_MAJORMINOR}" \
             -doMaf "${DO_MAF}" \
@@ -147,13 +148,12 @@ else
             -GL "${GT_LIKELIHOOD}" \
             -P "${N_CORES}" \
             -rf "${REGIONS}" \
-            -doPost "${DO_POST}"
+            -doPost "${DO_POST}")
     #   Are we missing a definiton for regions?
     elif [[ -z "${REGIONS}" ]]
     then
         echo "WRAPPER: $GROUP_2 sfs starting..." >&2
-        "${ANGSD_DIR}"/angsd \
-            -bam "${G2_SAMPLE_LIST}" \
+	WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
             -out "${OUT}"/"${GROUP_2}"_Intergenic \
             -doMajorMinor "${DO_MAJORMINOR}" \
             -doMaf "${DO_MAF}" \
@@ -169,12 +169,11 @@ else
             -ref "${REF_SEQ}" \
             -GL "${GT_LIKELIHOOD}" \
             -P "${N_CORES}" \
-            -doPost "${DO_POST}"
+            -doPost "${DO_POST}")
     #   Assuming a single reigon was defined in config file
     else
         echo "WRAPPER: $GROUP_2 sfs starting..." >&2
-        "${ANGSD_DIR}"/angsd \
-            -bam "${G2_SAMPLE_LIST}" \
+	WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
             -out "${OUT}"/"${GROUP_2}"_Intergenic \
             -doMajorMinor "${DO_MAJORMINOR}" \
             -doMaf "${DO_MAF}" \
@@ -190,9 +189,13 @@ else
             -ref "${REF_SEQ}" \
             -GL "${GT_LIKELIHOOD}" \
             -P "${N_CORES}" \
-            -r "${REGIONS}"
+            -r "${REGIONS}")
     fi
 fi
+# Check for advanced arguments, and overwrite any overlapping definitions
+FINAL_ARGS=$(source ${SOURCE}/Wrappers/Arg_Zipper.sh "${WRAPPER_ARGS}" "${ADVANCED_ARGS}")
+echo "Final arguments: ${FINAL_ARGS}" 1<&2
+"${ANGSD_DIR}"/angsd ${FINAL_ARGS}
 
 #   Estimate joint SFS using realSFS
 echo "WRAPPER: realSFS 2dsfs..." >&2

diff --git a/Wrappers/Genotypes.sh b/Wrappers/Genotypes.sh
@@ -39,8 +39,7 @@ ANGSD_DIR="${SOURCE}"/dependencies/angsd
 #   Do we have a regions file?
 if [[ -f "${REGIONS}" ]]
 then
-    "${ANGSD_DIR}"/angsd \
-        -bam "${SAMPLE_LIST}" \
+    WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
         -out "${OUT}"/"${PROJECT}"_snps \
         -indF "${SAMPLE_INBREEDING}" \
         -doMajorMinor "${DO_MAJORMINOR}" \
@@ -57,12 +56,11 @@ then
         -SNP_pval "${SNP_PVAL}" \
         -nInd "${N_IND}" \
         -minInd "${MIN_IND}" \
-        -P "${N_CORES}"
+        -P "${N_CORES}")
 #   Are we missing a definiton for regions?
 elif [[ -z "${REGIONS}" ]]
 then
-    "${ANGSD_DIR}"/angsd \
-        -bam "${SAMPLE_LIST}" \
+    WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
         -out "${OUT}"/"${PROJECT}"_snps \
         -indF "${SAMPLE_INBREEDING}" \
         -doMajorMinor "${DO_MAJORMINOR}" \
@@ -78,11 +76,10 @@ then
         -SNP_pval "${SNP_PVAL}" \
         -nInd "${N_IND}" \
         -minInd "${MIN_IND}" \
-        -P "${N_CORES}"
-#   Assuming a single reigon was defined in config file
+        -P "${N_CORES}")
+#   Assuming a single region was defined in config file
 else
-    "${ANGSD_DIR}"/angsd \
-        -bam "${SAMPLE_LIST}" \
+    WRAPPER_ARGS=$(echo -bam "${SAMPLE_LIST}" \
         -out "${OUT}"/"${PROJECT}"_snps \
         -indF "${SAMPLE_INBREEDING}" \
         -doMajorMinor "${DO_MAJORMINOR}" \
@@ -99,5 +96,9 @@ else
         -SNP_pval "${SNP_PVAL}" \
         -nInd "${N_IND}" \
         -minInd "${MIN_IND}" \
-        -P "${N_CORES}"
+        -P "${N_CORES}")
 fi
+# Check for advanced arguments, and overwrite any overlapping definitions
+FINAL_ARGS=$(source ${SOURCE}/Wrappers/Arg_Zipper.sh "${WRAPPER_ARGS}" "${ADVANCED_ARGS}")
+# echo "Final arguments: ${FINAL_ARGS}" 1<&2
+"${ANGSD_DIR}"/angsd ${FINAL_ARGS}