| 1 | #!/bin/bash -x |
|---|
| 2 | # renames output to 'treefile' |
|---|
| 3 | |
|---|
| 4 | usage() { |
|---|
| 5 | arb_message "arb_raxml called with wrong parameters (look @ console for errors)" |
|---|
| 6 | echo "Usage:" |
|---|
| 7 | echo " arb_raxml 'DNA' SEQFILE WEIGHTS TREE CONSTRAINT RANDOMSTART OPTIMIZEPARAMETERS SEARCH INITIALREARRANGEMENT SEED NUMBEROFRUNS TAKETREES [import|consense] RATEMODELNUC |
|---|
| 8 | NUMCATEGORIES" |
|---|
| 9 | echo " arb_raxml 'PROTEIN' SEQFILE WEIGHTS TREE CONSTRAINT RANDOMSTART OPTIMIZEPARAMETERS SEARCH INITIALREARRANGEMENT SEED NUMBEROFRUNS TAKETREES [import|consense] RATEMODELPROT MATRIXNAME" |
|---|
| 10 | exit 1 |
|---|
| 11 | } |
|---|
| 12 | |
|---|
| 13 | if [ -z "${13}" ]; then |
|---|
| 14 | usage |
|---|
| 15 | else |
|---|
| 16 | # parameters common for DNA and PROTEIN |
|---|
| 17 | DATA=$1 ; shift |
|---|
| 18 | SEQFILE=$1 ; shift |
|---|
| 19 | WEIGHTS=$1 ; shift |
|---|
| 20 | TREE=$1 ; shift |
|---|
| 21 | CONSTRAINT=$1 ; shift |
|---|
| 22 | |
|---|
| 23 | RANDOMSTART=$1 ; shift |
|---|
| 24 | OPTIMIZEPARAMETERS=$1 ; shift |
|---|
| 25 | SEARCH=$1 ; shift |
|---|
| 26 | INITIALREARRANGEMENT=$1 ; shift |
|---|
| 27 | SEED=$1 ; shift |
|---|
| 28 | |
|---|
| 29 | NUMBEROFRUNS=$1 ; shift |
|---|
| 30 | TAKETREES=$1 ; shift |
|---|
| 31 | CONSENSE=$1 ; shift |
|---|
| 32 | |
|---|
| 33 | if [ -z "$CONSENSE" ]; then |
|---|
| 34 | usage |
|---|
| 35 | fi |
|---|
| 36 | |
|---|
| 37 | if [ "$DATA" == "DNA" ]; then |
|---|
| 38 | if [ -z "$1" ]; then |
|---|
| 39 | usage |
|---|
| 40 | fi |
|---|
| 41 | RATEMODELNUC=$1 |
|---|
| 42 | NUMCATEGORIES=$2 |
|---|
| 43 | else |
|---|
| 44 | if [ "$DATA" == "PROTEIN" ]; then |
|---|
| 45 | # if [ -z "$2" ]; then |
|---|
| 46 | # usage |
|---|
| 47 | # fi |
|---|
| 48 | RATEMODELPROT=$1 |
|---|
| 49 | MATRIXNAME=$2 |
|---|
| 50 | else |
|---|
| 51 | arb_message "Unknown datatype '$DATA'" |
|---|
| 52 | exit 1 |
|---|
| 53 | fi |
|---|
| 54 | fi |
|---|
| 55 | |
|---|
| 56 | if [ -z "$SEED" ]; then |
|---|
| 57 | # seconds since 1970 |
|---|
| 58 | SEED=`date +%s` |
|---|
| 59 | fi |
|---|
| 60 | |
|---|
| 61 | # check inputfiles |
|---|
| 62 | if [ \! -s $SEQFILE ]; then |
|---|
| 63 | arb_message "Missing or empty sequence file '$SEQFILE'" |
|---|
| 64 | exit 1 |
|---|
| 65 | fi |
|---|
| 66 | if [ \! -s $WEIGHTS ]; then |
|---|
| 67 | arb_message "Missing or empty weights file '$WEIGHTS'" |
|---|
| 68 | exit 1 |
|---|
| 69 | fi |
|---|
| 70 | |
|---|
| 71 | # generate tree file |
|---|
| 72 | TREEPARAMS= |
|---|
| 73 | HAVEINPUTTREE=0 |
|---|
| 74 | if [ "$TREE" != "????" -a ! -z "$TREE" ]; then |
|---|
| 75 | CMD="arb_export_tree $TREE" |
|---|
| 76 | echo "$CMD" |
|---|
| 77 | echo `$CMD > treefile.in` |
|---|
| 78 | if [ \! -s treefile.in ]; then |
|---|
| 79 | arb_message "Couldn't export tree '$TREE'" |
|---|
| 80 | exit 1 |
|---|
| 81 | fi |
|---|
| 82 | if [ "$CONSTRAINT" == "1" ]; then |
|---|
| 83 | TREEPARAMS="-r treefile.in" |
|---|
| 84 | else |
|---|
| 85 | TREEPARAMS="-t treefile.in" |
|---|
| 86 | fi |
|---|
| 87 | HAVEINPUTTREE=1 |
|---|
| 88 | else |
|---|
| 89 | if [ "$RANDOMSTART" == "1" ]; then |
|---|
| 90 | TREEPARAMS=-d |
|---|
| 91 | fi |
|---|
| 92 | # otherwise use default parsimony tree |
|---|
| 93 | fi |
|---|
| 94 | |
|---|
| 95 | # model dependent parameters |
|---|
| 96 | DEPENDENT_PARAMS= |
|---|
| 97 | if [ "$OPTIMIZEPARAMETERS" == "1" ]; then |
|---|
| 98 | if [ "$DATA" == "DNA" ]; then |
|---|
| 99 | case "$RATEMODELNUC" in |
|---|
| 100 | GTRMIX | GTRGAMMA ) |
|---|
| 101 | DEPENDENT_PARAMS="$DEPENDENT_PARAMS -k" |
|---|
| 102 | ;; |
|---|
| 103 | * ) |
|---|
| 104 | arb_message "Ignored 'Optimize branches/parameters'\n(only usable with GTRMIX or GTRGAMMA)" |
|---|
| 105 | ;; |
|---|
| 106 | esac |
|---|
| 107 | else |
|---|
| 108 | case "$RATEMODELPROT" in |
|---|
| 109 | "PROTMIX" | "PROTGAMMA" ) |
|---|
| 110 | DEPENDENT_PARAMS="$DEPENDENT_PARAMS -k" |
|---|
| 111 | ;; |
|---|
| 112 | * ) |
|---|
| 113 | arb_message "Ignored 'Optimize branches/parameters'\n(only usable with PROTMIX or PROTGAMMA)" |
|---|
| 114 | ;; |
|---|
| 115 | esac |
|---|
| 116 | fi |
|---|
| 117 | fi |
|---|
| 118 | |
|---|
| 119 | if [ "$DATA" == "DNA" ]; then |
|---|
| 120 | if [ "$RATEMODELNUC" == "GTRCAT" ]; then |
|---|
| 121 | DEPENDENT_PARAMS="$DEPENDENT_PARAMS -c $NUMCATEGORIES" |
|---|
| 122 | fi |
|---|
| 123 | fi |
|---|
| 124 | |
|---|
| 125 | # search algorithm |
|---|
| 126 | NEEDINPUTTREE=0 |
|---|
| 127 | USESINPUTTREE=0 |
|---|
| 128 | NEEDGAMMA=0 |
|---|
| 129 | GENERATEDTREES=$NUMBEROFRUNS |
|---|
| 130 | |
|---|
| 131 | case "$SEARCH" in |
|---|
| 132 | "e" ) |
|---|
| 133 | NEEDINPUTTREE=1 |
|---|
| 134 | NEEDGAMMA=1 |
|---|
| 135 | ;; |
|---|
| 136 | "t" ) |
|---|
| 137 | NEEDINPUTTREE=1 |
|---|
| 138 | ;; |
|---|
| 139 | "d" ) # modes that use (but don't expect) an inputtree |
|---|
| 140 | USESINPUTTREE=1 |
|---|
| 141 | ;; |
|---|
| 142 | |
|---|
| 143 | "a" ) # rapid bootstrap analysis |
|---|
| 144 | DEPENDENT_PARAMS="$DEPENDENT_PARAMS -x $SEED" |
|---|
| 145 | GENERATEDTREES=1 |
|---|
| 146 | if [ "$TAKETREES" != "1" ]; then |
|---|
| 147 | arb_message "'rapid bootstrap analysis' only creates 1 tree\n(can't select $TAKETREES trees, falling back to 1)" |
|---|
| 148 | TAKETREES=1 |
|---|
| 149 | fi |
|---|
| 150 | ;; |
|---|
| 151 | "p" ) # add new sequences (MP) |
|---|
| 152 | NEEDINPUTTREE=1 |
|---|
| 153 | GENERATEDTREES=1 |
|---|
| 154 | if [ "$NUMBEROFRUNS" != "1" ]; then |
|---|
| 155 | arb_message "Ignoring number of runs (only performing 1 run)" |
|---|
| 156 | NUMBEROFRUNS=1 |
|---|
| 157 | fi |
|---|
| 158 | TAKETREES=1 |
|---|
| 159 | ;; |
|---|
| 160 | esac |
|---|
| 161 | |
|---|
| 162 | if [ "$NEEDINPUTTREE" != "$HAVEINPUTTREE" ]; then |
|---|
| 163 | if [ "$NEEDINPUTTREE" == "1" ]; then |
|---|
| 164 | arb_message "Please select an input tree" |
|---|
| 165 | exit 1; |
|---|
| 166 | else |
|---|
| 167 | if [ "$USESINPUTTREE" == "0" ]; then |
|---|
| 168 | arb_message "Given input tree ignored" |
|---|
| 169 | # dont use tree: |
|---|
| 170 | TREEPARAMS= |
|---|
| 171 | fi |
|---|
| 172 | fi |
|---|
| 173 | fi |
|---|
| 174 | |
|---|
| 175 | if [ "$NEEDGAMMA" == "1" ]; then |
|---|
| 176 | if [ "$DATA" == "DNA" ]; then |
|---|
| 177 | case "$RATEMODELNUC" in |
|---|
| 178 | "GTRGAMMA" | "GTRGAMMAI" ) |
|---|
| 179 | ;; |
|---|
| 180 | * ) |
|---|
| 181 | arb_message "Please choose GAMMA substitution model" |
|---|
| 182 | exit 1 |
|---|
| 183 | ;; |
|---|
| 184 | esac |
|---|
| 185 | else |
|---|
| 186 | case "$RATEMODELPROT" in |
|---|
| 187 | "PROTGAMMA" | "PROTGAMMAI" ) |
|---|
| 188 | ;; |
|---|
| 189 | * ) |
|---|
| 190 | arb_message "Please choose GAMMA rate model" |
|---|
| 191 | exit 1 |
|---|
| 192 | ;; |
|---|
| 193 | esac |
|---|
| 194 | fi |
|---|
| 195 | fi |
|---|
| 196 | |
|---|
| 197 | MODELPARAMS= |
|---|
| 198 | if [ "$DATA" == "DNA" ]; then |
|---|
| 199 | MODELPARAMS="-m $RATEMODELNUC" |
|---|
| 200 | else |
|---|
| 201 | MODELPARAMS="-m $RATEMODELPROT$MATRIXNAME" |
|---|
| 202 | fi |
|---|
| 203 | |
|---|
| 204 | # build parameters for raxml |
|---|
| 205 | RAXMLPARAMS="$MODELPARAMS -f $SEARCH" |
|---|
| 206 | if [ ! -z "$INITIALREARRANGEMENT" ]; then |
|---|
| 207 | RAXMLPARAMS="$RAXMLPARAMS -i $INITIALREARRANGEMENT" |
|---|
| 208 | # otherwise autodetect by RAxML |
|---|
| 209 | fi |
|---|
| 210 | if [ "$NUMBEROFRUNS" != "1" ]; then |
|---|
| 211 | RAXMLPARAMS="$RAXMLPARAMS -# $NUMBEROFRUNS" |
|---|
| 212 | fi |
|---|
| 213 | RAXMLPARAMS="$RAXMLPARAMS -s $SEQFILE -a $WEIGHTS $TREEPARAMS $DEPENDENT_PARAMS -n ThisRun" |
|---|
| 214 | |
|---|
| 215 | echo "------------------------------------------" |
|---|
| 216 | echo "Calling raxmlHPC $RAXMLPARAMS" |
|---|
| 217 | time raxmlHPC $RAXMLPARAMS |
|---|
| 218 | echo "------------------------------------------" |
|---|
| 219 | |
|---|
| 220 | # ../PERL_SCRIPTS/ARBTOOLS/raxml2arb.pl |
|---|
| 221 | raxml2arb.pl ThisRun "$GENERATEDTREES" "$TAKETREES" "$CONSENSE" |
|---|
| 222 | fi |
|---|