用于系统发育树的循环

时间:2017-02-17 20:40:41

标签: for-loop multiprocessing slurm

我正在使用一种名为RAxML的系统发育软件,我想为每个phylip文件构建单个树。对于包含三个phylip文件的目录,我执行了以下操作

 ##files in directory
 Ortho1.phy Ortho6.Phy Ortho6.Phy

for f in /home/Single_trees/trimmed_alignment/*.phy; do raxmlHPC -f a -x 100 -m PROTGAMMAAUTO -p 100 -s $f -N 100 -n $f.tree; done;

但这总是给我一个错误,说不允许$符号。

raxmlHPC: axml.c:5236: analyzeRunId: Assertion `0' failed.

错误字符/运行ID中不允许

有更好的方法吗?我在https://rc.fas.harvard.edu/resources/documentation/submitting-large-numbers-of-jobs-to-odyssey/使用此链接尝试将作业数组用于非连续命名文件,但无法实现它。

这是我为阵列作业提交尝试的内容:

 #!/bin/bash -l
 #
 # raxml.sbatch
 #
 #SBATCH -J consensus       # A single job name for the array
 #SBATCH -p high # best partition for single core small jobs
 #SBATCH -n 12              # one core
 #SBATCH -N 1              # on one node
 #SBATCH -t 100:00:00         # Running time of 2 hours
 #SBATCH --mem 18000        # Memory request of 4 GB
 #SBATCH -o raxml_%A_%a.out # Standard output
 #SBATCH -e raxml_%A_%a.err # Standard error
 module load raxml

  for FILES in /home/aligned_fasta/.phy; do
   echo ${FILES}
  done;

  # grab out filename from the array exported from our 'parent' shell
  FILENAME=${FILES[$SLURM_ARRAY_TASK_ID]}

  # make & move into new directory, and run!
 mkdir ${FILENAME}_out
 cd ${FILENAME}_out
 raxmlHPC -f a -x 100 -m PROTGAMMAAUTO -p 100  -s $FILENAME -N 100 -n $FILENAME.tree 

 #Now, we grab all the appropriate files and submit them en-batch with an  array:
 # grab the files, and export it so the 'child' sbatch jobs can access it

导出FILES =($(ls -1 .phy))

 # get size of array
 NUMPHY=${#FILES[@]}
 # now subtract 1 as we have to use zero-based indexing (first cell is 0)
 ZBNUMPHY=$(($NUMPHY - 1))

 # now submit to SLURM
 if [ $ZBNUMPHY -ge 0 ]; then
 sbatch --array=0-$ZBNUMPHY raxml.sbatch 
 fi

我使用sbatch -array = 0-10 raxml.sh提交但是它没有用。

2 个答案:

答案 0 :(得分:0)

刚想出来的东西。基本上,我将重命名文件,使它们是连续的,我可以使用slurm。

 ls *.phy | cat -n | while read num file; do mv $file ${file/./.$num.}; done

所以文件将是

 Ortho1.1.phy Ortho6.2.Phy Ortho6.3.Phy

然后你可以通过以下方式完成:

#!/bin/bash -l
# SBATCH -J tree

###### Standard out and Standard Error output files with the job number in the name.
#SBATCH -o tre_%A.%a.out
#SBATCH -e tre_%A.%a.err

###### number of nodes
###SBATCH --nodes=6
###SBATCH --nodes=6

###### number of processors
#SBATCH -n 16
###SBATCH --cpus-per-task=4

###### Spread the tasks evenly among the nodes
####BATCH --ntasks-per-node=8

###### coupled with array
####SBATCH --ntasks=1-179

#SBATCH --time=300:00:00

#SBATCH -p high

#SBATCH --mem 24000 

###### Want the node exclusively
### SBATCH --exclusive

#SBATCH --array=1-3

module load raxml

for i in $SLURM_ARRAY_TASK_ID.phy
do
echo $i
done

### tree
 raxmlHPC-PTHREADS -f a -x 100 -m PROTGAMMAAUTO -p 100 -T 16 -s $i -N 100 -n $i.tree 

这将为您提供可用于构建共识树的单独树。

答案 1 :(得分:0)

我认为您可能正在引导并获得共识系统发育,在这种情况下,RAxML内有一种特殊情况。我一有空就把它张贴在这里。