Search zinc22.py

Description

usage: init_partitioned_search.py [-i] input_file [-o] results_out [--get-vendors] [-s] 

search for smiles and vendor codes by zinc22 id

required arguments:
  input_file            file containing list of zinc ids or vendor codes to look up
  results_out           destination for output directory

optional arguments:
  -h, --help            show this help message and exit
  --vendor-search       look up molecules by vendor code instead of zinc id
  --get-vendors         get vendor supplier codes associated with zinc id
  -s                    submit slurm jobs and start the search process, can be omitted if you are only looking to split a list of molecules into tranches

init_partitioned_search.py is a script that replaces search_zinc22.py for looking up zinc ids on the zinc22 system in a more efficient manner. The operation is simple- provide a file containing a list of zincids and the script will split the input into tranches for search across all zinc22 databases.

The output format is as follows: SMILES ZINC_ID TRANCHE_NAME

With --get-vendors or --vendor-search the output format looks like this:

SMILES ZINC_ID VENDOR_ID TRANCHE_NAME CATALOG</nowiki>

Meaning the script will find all vendor information and smiles associated with the provided zinc ids or vendor codes.

Location

You can activate the environment, using

 source /nfs/soft/zinc22/search_zinc/miniconda/bin/activate zinc22_search 

Side note about slow queries

Depending on the molecules you happen to be looking up, your search may go by very quickly, or somewhat slowly. Smaller molecules tend to look up very quickly, while larger molecules take longer to find. We're working on it.

Tracking Progress

After submitting the search, you can look at the slurm queue to see the status of your job. Jobs will be labeled 'search_zinc22'

Using Output

When all jobs are done, the output folder will have input files, log files, and .results files. If results files are missing for a tranche, refer to the logs to learn more or contact the JJI team for troubleshooting. You can merge all results together in a file with a simple command

 cat *.results > yourresultsfile.txt

Usage w/ Bash on BKS cluster

source /nfs/soft/zinc22/search_zinc/miniconda/bin/activate /nfs/soft/zinc22/search_zinc/miniconda/envs/zinc22_search
python /nfs/soft/zinc22/search_zinc/init_partitioned_search.py -i /path/to/input/file -o /path/to/output/folder -s
python /nfs/soft/zinc22/search_zinc/init_partitioned_search.py -i /path/to/input/file -o /path/to/output/folder -s --get_vendors
 

Usage w/ Csh on BKS cluster

This version of the tool is currently only compatible with Bash.

Dealing with NULL

Sometimes a ZINC ID will fail to look up. This could be because a server is down (the script will notify you if this is the case), or because the ID is missing from the system for some reason. In this case, it may be helpful to separate the molecules that didn't look up from the molecules that did. You may want to save them for later when the servers come back online, or to run a deeper search with comb_legacy_files.py (more on this below).

How to:

[env]$ cat legitimate_ids.txt > input.txt
[env]$ echo ZINCzz00ZZZZZZZZ >> input.txt
[env]$ echo ZINCyy00AAAAAAAA >> input.txt
[env]$ echo ZINCxx00BBBBBBBB >> input.txt
[env]$ python search_zinc.py input.txt output.txt
Searching Zinc22:  |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX| 100.0%  0.00s 23/23 complete!
[env]$ grep "_null_" output.txt
_null_ ZINCzz00ZZZZZZZZ H33P270
_null_ ZINCyy00AAAAAAAA H34P280
_null_ ZINCxx00BBBBBBBB H35P290

search_zinc.py will not omit IDs that don't look up from the output, instead it will return the zinc id with "_null_" in every other field. Therefore we can use grep to filter our results.

[env]$ grep "_null_" output.txt > missing.txt
[env]$ grep -v "_null_" output.txt > found.txt

It should be very infrequent that ZINC IDs don't look up, but if this happens you can use the following script:

comb_legacy_files.py

python3 /mnt/nfs/home/xyz/btingle/bin/2dload.testing/utils-2d/tin/misc/comb_legacy_files.py [INPUT_ZINC_IDS_FILE]

You don't need to source any particular python 3 environment for this script, but the environment used for search_zinc22.py will work just fine here.

This script will comb through our deprecated files and attempt to locate your ZINC IDs there. This script will create a file called "result" in your current directory containing all the smiles found.

If you're looking from vendor information, you can look up the SMILES you get back in arthor/smallworld sets to find vendor codes. Functionality is planned in search_zinc22.py for looking up by SMILES, but not implemented yet.

If after this you're STILL unable to find your zinc ids, you can send them to our development team and we will find them for you.

Email ben@tingle.org, ccing khtang015@gmail.com and josecastanon4@gmail.com. Include your missing file as an attachment.