Search code examples
bashawksedcat

cat/sed: concatenation of the fragments of the log filles

I have a folder consisted of many logs. Each log have a similar format.

This is the log1

Finding intermodel H-bonds
Finding intramodel H-bonds
Constraints relaxed by 0.55 angstroms and 40 degrees
Models used:
    1.1 SarsCov2_Y6A_nsp5holo_rep1.pdb

6 H-bonds
H-bonds (donor, acceptor, hydrogen, D..A dist, D-H..A dist):
/? SER 144 OG   /d UNL 1 S     /? SER 144 HG    3.940  3.529
/? HIS 163 NE2  /d UNL 1 S     no hydrogen      3.821  N/A
/? GLN 189 NE2  /d UNL 1 O     /? GLN 189 1HE2  3.178  2.453
/d UNL 1 N      /? THR 25 OG1  /d UNL 1 HN      2.755  2.270
/d UNL 1 N      /? CYS 44 O    /d UNL 1 HN      3.277  2.501
/d UNL 1 N      /? ARG 188 O   /d UNL 1 HN      3.056  2.055

log2

Finding intermodel H-bonds
Finding intramodel H-bonds
Constraints relaxed by 0.55 angstroms and 40 degrees
Models used:
    1.1 SarsCov2_06I_nsp5holo_rep1.pdb    
4 H-bonds
H-bonds (donor, acceptor, hydrogen, D..A dist, D-H..A dist):
/? THR 26 N     /d UNL 1 O      /? THR 26 H      3.579  2.754
/? ASN 142 ND2  /d UNL 1 O      /? ASN 142 1HD2  3.250  2.324
/d UNL 1 N      /? THR 26 O     /d UNL 1 H       3.458  2.630
/d UNL 1 N      /? HIS 163 NE2  /d UNL 1 HN      3.222  2.456

This is the log 3:

Finding intermodel H-bonds
Finding intramodel H-bonds
Constraints relaxed by 0.55 angstroms and 40 degrees
Models used:
    1.1 SarsCov2_X7V_nsp5holo_rep1.pdb

2 H-bonds
H-bonds (donor, acceptor, hydrogen, D..A dist, D-H..A dist):
/? GLN 189 NE2  /d UNL 1 O    /? GLN 189 1HE2  3.185  2.258
/d UNL 1 N      /? LEU 141 O  /d UNL 1 HN      2.868  1.958

I need to fuse all the logs together taking only the strings starting from # H-bonds adding the name of the initial file in the same line:

This is fused log produced by combining log1 -log 3:

log 1: 6 H-bonds
H-bonds (donor, acceptor, hydrogen, D..A dist, D-H..A dist):
/? SER 144 OG   /d UNL 1 S     /? SER 144 HG    3.940  3.529
/? HIS 163 NE2  /d UNL 1 S     no hydrogen      3.821  N/A
/? GLN 189 NE2  /d UNL 1 O     /? GLN 189 1HE2  3.178  2.453
/d UNL 1 N      /? THR 25 OG1  /d UNL 1 HN      2.755  2.270
/d UNL 1 N      /? CYS 44 O    /d UNL 1 HN      3.277  2.501
/d UNL 1 N      /? ARG 188 O   /d UNL 1 HN      3.056  2.055


log 2: 4 H-bonds
H-bonds (donor, acceptor, hydrogen, D..A dist, D-H..A dist):
/? THR 26 N     /d UNL 1 O      /? THR 26 H      3.579  2.754
/? ASN 142 ND2  /d UNL 1 O      /? ASN 142 1HD2  3.250  2.324
/d UNL 1 N      /? THR 26 O     /d UNL 1 H       3.458  2.630
/d UNL 1 N      /? HIS 163 NE2  /d UNL 1 HN      3.222  2.456

log3: 2 H-bonds
H-bonds (donor, acceptor, hydrogen, D..A dist, D-H..A dist):
/? GLN 189 NE2  /d UNL 1 O    /? GLN 189 1HE2  3.185  2.258
/d UNL 1 N      /? LEU 141 O  /d UNL 1 HN      2.868  1.958

I've tried trivial solution with CAT but it does not work correcly since in each log I have different number of lines and TAIL could not recognize it correctly:

for log in ${results}/*_rep"${i}".log; do
  log_name=$(basename "$log" .log)
  echo "$log_name" >> ${results}/combined.log
  cat $log | tail -n 10 >> ${results}/combined.log
done

may I use cat in some specific expressin to recognize lines or alternatively I have to use SED before CAT to delete unused lines from each initial log ??

Solution

  • This awk does the job:

    awk '
FNR==1 {p=0}
/^[0-9]+[[:space:]]+H-bonds$/ && FNR!=NR {printf "\n"}
/^[0-9]+[[:space:]]+H-bonds$/ {printf "log %d: ", ++c; p=1}
p==1'

    For each given file:

    • Stop printing when each new file starts.
    • If a line in the file matches a pattern for 4 H-bonds etc, print an empty line break, if it's not the first file. Then print the log number, and set the flag p to begin printing that log file.
    • Note that instead of testing for the same regex twice, you could omit the first one, and put if (FNR!=NR) {printf "\n"} inside the block of the second one. That's mainly about readability.
    • I'm unsure which pattern expression you need for the file names, *.log is an example. Maybe "${results}"/*_rep*.log?