6662 |
01 Apr 22 |
nicklas |
#!/bin/bash |
6662 |
01 Apr 22 |
nicklas |
2 |
## |
6662 |
01 Apr 22 |
nicklas |
## Pipeline script for checking status of NextSeq sequencing run. |
6662 |
01 Apr 22 |
nicklas |
4 |
## |
6662 |
01 Apr 22 |
nicklas |
## Environment variables that should be defined before calling this script |
6662 |
01 Apr 22 |
nicklas |
## -AllRunArchives: White-space separated list of locations to search for sequencing data |
6662 |
01 Apr 22 |
nicklas |
## -BARCODE: Barcode of the flow cell that we want to find information for |
6662 |
01 Apr 22 |
nicklas |
8 |
## |
6662 |
01 Apr 22 |
nicklas |
## The output is a number of key-value pairs. All values may not be present. |
6662 |
01 Apr 22 |
nicklas |
10 |
## |
6662 |
01 Apr 22 |
nicklas |
## -RunArchive: The path to the data folder for the flow cell |
6662 |
01 Apr 22 |
nicklas |
## -Config: Date and time the 'Config' folder was last modified |
6662 |
01 Apr 22 |
nicklas |
## -RunParameters: Date and time the 'RunParameters.xml' file was last modified |
6662 |
01 Apr 22 |
nicklas |
## -BgzfCount: Number of files ending with 'bgzf.bci' |
6662 |
01 Apr 22 |
nicklas |
## -NumLanes: Value from <NumLanes> tag in RunParameters.xml |
6662 |
01 Apr 22 |
nicklas |
## -Read1: Value from <Read1> tag in RunParameters.xml |
6662 |
01 Apr 22 |
nicklas |
## -Read2: Value from <Read2> tag in RunParameters.xml |
6662 |
01 Apr 22 |
nicklas |
## -Index1Read: Value from <Index1Read> tag in RunParameters.xml |
6662 |
01 Apr 22 |
nicklas |
## -Index2Read: Value from <Index2Read> tag in RunParameters.xml |
6662 |
01 Apr 22 |
nicklas |
## -NextSeqSerial: Value from <InstrumentID> tag in RunParameters.xml |
6662 |
01 Apr 22 |
nicklas |
## -RunCompletionStatus: Date and time the 'RunCompletionStatus.xml' was last modified |
6662 |
01 Apr 22 |
nicklas |
22 |
|
6662 |
01 Apr 22 |
nicklas |
# Format string for file dates/times |
6662 |
01 Apr 22 |
nicklas |
24 |
DATE_FORMAT="%Y%m%d %H%M%S" |
6662 |
01 Apr 22 |
nicklas |
25 |
|
6662 |
01 Apr 22 |
nicklas |
# Try to find a folder inside run-archive that has the barcode in the name |
6662 |
01 Apr 22 |
nicklas |
# The folder may not yet exist so a missing folder is not an error |
6663 |
01 Apr 22 |
nicklas |
28 |
DATA_FOLDER=`find ${AllRunArchives} -maxdepth 2 -iname "*${BARCODE}*" -type d -print 2> /dev/null || true` |
6662 |
01 Apr 22 |
nicklas |
29 |
|
6662 |
01 Apr 22 |
nicklas |
# Fail if more than one folder is found |
6662 |
01 Apr 22 |
nicklas |
31 |
readarray -t lines <<< "${DATA_FOLDER}" |
6662 |
01 Apr 22 |
nicklas |
32 |
if [ ! ${#lines[@]} -eq 1 ]; then |
6662 |
01 Apr 22 |
nicklas |
33 |
echo "Found ${#lines[@]} data folders for flow cell ${BARCODE}" 1>&2 |
6662 |
01 Apr 22 |
nicklas |
34 |
echo ${DATA_FOLDER} 1>&2 |
6662 |
01 Apr 22 |
nicklas |
35 |
exit 1 |
6662 |
01 Apr 22 |
nicklas |
36 |
fi |
6662 |
01 Apr 22 |
nicklas |
37 |
|
6662 |
01 Apr 22 |
nicklas |
38 |
echo RunArchive: ${DATA_FOLDER} |
6662 |
01 Apr 22 |
nicklas |
# Config folder is created immediately when starting the NextSeq |
6662 |
01 Apr 22 |
nicklas |
# We use the date of this folder to set the start date of the job |
6662 |
01 Apr 22 |
nicklas |
41 |
if [ -d "${DATA_FOLDER}/Config" ]; then |
6662 |
01 Apr 22 |
nicklas |
42 |
echo "Config: `date +"${DATE_FORMAT}" -r "${DATA_FOLDER}/Config"`" |
6662 |
01 Apr 22 |
nicklas |
43 |
fi |
6662 |
01 Apr 22 |
nicklas |
44 |
|
6662 |
01 Apr 22 |
nicklas |
# RunParameters.xml is created after clustering |
6662 |
01 Apr 22 |
nicklas |
# We extract information about number of reads and lanes |
6662 |
01 Apr 22 |
nicklas |
# and compare that to the number of *.bgzf.bci files we can find |
6662 |
01 Apr 22 |
nicklas |
# This gives an estimate of the current sequencing cycle and we can |
6662 |
01 Apr 22 |
nicklas |
# use this for progress reporting |
6662 |
01 Apr 22 |
nicklas |
50 |
RUN_PARAMETERS=${DATA_FOLDER}/RunParameters.xml |
6662 |
01 Apr 22 |
nicklas |
51 |
if [ -f "${RUN_PARAMETERS}" ]; then |
6662 |
01 Apr 22 |
nicklas |
52 |
echo "RunParameters: `date +"${DATE_FORMAT}" -r "${RUN_PARAMETERS}"`" |
6662 |
01 Apr 22 |
nicklas |
53 |
echo "BgzfCount: `find "${DATA_FOLDER}" -type f -name *.bgzf.bci | wc -l`" |
6662 |
01 Apr 22 |
nicklas |
54 |
echo "NumLanes: `grep '<NumLanes>' "${RUN_PARAMETERS}" | cut -d '>' -f 2 | cut -d '<' -f 1`" |
6662 |
01 Apr 22 |
nicklas |
55 |
echo "Read1: `grep '<Read1>' "${RUN_PARAMETERS}" | cut -d '>' -f 2 | cut -d '<' -f 1`" |
6662 |
01 Apr 22 |
nicklas |
56 |
echo "Read2: `grep '<Read2>' "${RUN_PARAMETERS}" | cut -d '>' -f 2 | cut -d '<' -f 1`" |
6662 |
01 Apr 22 |
nicklas |
57 |
echo "Index1Read: `grep '<Index1Read>' "${RUN_PARAMETERS}" | cut -d '>' -f 2 | cut -d '<' -f 1`" |
6662 |
01 Apr 22 |
nicklas |
58 |
echo "Index2Read: `grep '<Index2Read>' "${RUN_PARAMETERS}" | cut -d '>' -f 2 | cut -d '<' -f 1`" |
6662 |
01 Apr 22 |
nicklas |
59 |
echo "NextSeqSerial: `grep '<InstrumentID>' "${RUN_PARAMETERS}" | cut -d '>' -f 2 | cut -d '<' -f 1`" |
6662 |
01 Apr 22 |
nicklas |
60 |
fi |
6662 |
01 Apr 22 |
nicklas |
61 |
|
6662 |
01 Apr 22 |
nicklas |
# RunCompletionStatus.xml is created when everything is complete |
6662 |
01 Apr 22 |
nicklas |
# This becomes the end date of the job and should trigger |
6662 |
01 Apr 22 |
nicklas |
# Reggie to start file checks and secondary analysis |
6662 |
01 Apr 22 |
nicklas |
65 |
if [ -f "${DATA_FOLDER}/RunCompletionStatus.xml" ]; then |
6662 |
01 Apr 22 |
nicklas |
66 |
echo "RunCompletionStatus: `date +"${DATE_FORMAT}" -r "${DATA_FOLDER}/RunCompletionStatus.xml"`" |
6662 |
01 Apr 22 |
nicklas |
67 |
fi |