Louis BECQUEY

Rethink how inferred mappings work

This diff is collapsed. Click to expand it.
...@@ -12,7 +12,6 @@ done ...@@ -12,7 +12,6 @@ done
12 12
13 PROCESS_TO_KILL="statistics.py" 13 PROCESS_TO_KILL="statistics.py"
14 PROCESS_LIST=`ps ax | grep -Ei ${PROCESS_TO_KILL} | grep -Eiv '(grep|vi statistics.py)' | awk ' { print $1;}'` 14 PROCESS_LIST=`ps ax | grep -Ei ${PROCESS_TO_KILL} | grep -Eiv '(grep|vi statistics.py)' | awk ' { print $1;}'`
15 -KILLED=
16 for KILLPID in $PROCESS_LIST; do 15 for KILLPID in $PROCESS_LIST; do
17 if [ ! -z $KILLPID ];then 16 if [ ! -z $KILLPID ];then
18 kill -9 $KILLPID 17 kill -9 $KILLPID
......
...@@ -3,8 +3,8 @@ import subprocess, os, sys ...@@ -3,8 +3,8 @@ import subprocess, os, sys
3 3
4 # Put a list of problematic chains here, they will be properly deleted and recomputed 4 # Put a list of problematic chains here, they will be properly deleted and recomputed
5 problems = [ 5 problems = [
6 -"4v9n_1_DA_1-2879", 6 + "1k73_1_A",
7 -"4v9n_1_DA_148-2875" 7 + "1k73_1_B"
8 ] 8 ]
9 9
10 path_to_3D_data = sys.argv[1] 10 path_to_3D_data = sys.argv[1]
...@@ -15,6 +15,7 @@ for p in problems: ...@@ -15,6 +15,7 @@ for p in problems:
15 print() 15 print()
16 print() 16 print()
17 print() 17 print()
18 + homology = ('-' in p)
18 19
19 # Remove the datapoints files and 3D files 20 # Remove the datapoints files and 3D files
20 subprocess.run(["rm", '-f', path_to_3D_data + f"/rna_mapped_to_Rfam/{p}.cif"]) 21 subprocess.run(["rm", '-f', path_to_3D_data + f"/rna_mapped_to_Rfam/{p}.cif"])
...@@ -25,16 +26,25 @@ for p in problems: ...@@ -25,16 +26,25 @@ for p in problems:
25 # Find more information 26 # Find more information
26 structure = p.split('_')[0] 27 structure = p.split('_')[0]
27 chain = p.split('_')[2] 28 chain = p.split('_')[2]
28 - families = [ f.split('.')[1] for f in files ] # The RFAM families this chain has been mapped onto 29 + if homology:
29 - 30 + families = [ f.split('.')[1] for f in files ] # The RFAM families this chain has been mapped onto
30 - # Delete the chain from the database, and the associated nucleotides and re_mappings, using foreign keys 31 +
31 - for fam in families: 32 + # Delete the chain from the database, and the associated nucleotides and re_mappings, using foreign keys
32 - command = ["sqlite3", "results/RNANet.db", f"PRAGMA foreign_keys=ON; delete from chain where structure_id=\"{structure}\" and chain_name=\"{chain}\" and rfam_acc=\"{fam}\";"] 33 + for fam in families:
34 + command = ["sqlite3", "results/RNANet.db", f"PRAGMA foreign_keys=ON; delete from chain where structure_id=\"{structure}\" and chain_name=\"{chain}\" and rfam_acc=\"{fam}\";"]
35 + print(' '.join(command))
36 + subprocess.run(command)
37 +
38 + command = ["python3.8", "RNAnet.py", "--3d-folder", path_to_3D_data, "--seq-folder", path_to_seq_data, "-r", "20.0", "--extract", "--only", p]
39 + else:
40 + # Delete the chain from the database, and the associated nucleotides and re_mappings, using foreign keys
41 + command = ["sqlite3", "results/RNANet.db", f"PRAGMA foreign_keys=ON; delete from chain where structure_id=\"{structure}\" and chain_name=\"{chain}\" and rfam_acc is null;"]
33 print(' '.join(command)) 42 print(' '.join(command))
34 subprocess.run(command) 43 subprocess.run(command)
35 44
45 + command = ["python3.8", "RNAnet.py", "--3d-folder", path_to_3D_data, "--seq-folder", path_to_seq_data, "-r", "20.0", "--no-homology", "--extract", "--only", p]
46 +
36 # Re-run RNANet 47 # Re-run RNANet
37 - command = ["python3.8", "RNAnet.py", "--3d-folder", path_to_3D_data, "--seq-folder", path_to_seq_data, "-r", "20.0", "--extract", "--only", p]
38 print('\n',' '.join(command),'\n') 48 print('\n',' '.join(command),'\n')
39 subprocess.run(command) 49 subprocess.run(command)
40 50
......
This diff is collapsed. Click to expand it.