BROOD generates seven output files in addition to the hitlist. All of these files begin with the prefix specified by the -prefix flag, that by default is “brood”. The also include an integer to retain unique file names in the working directory (in this example, we will use the integer “1”). These files include:
- brood_1.info, the info file
- brood_1.log, the log file
- brood_1.param, the param file
- brood_1.rpt, the report file
- brood_1.csv, the spreadsheet file
By default, there is one hitlist and the contents of the hitlist will be determined by the command flags.
In post-processing of the search, some fragments are removed from the hitlist. These are preserved in the “removed” file.
If the -ET flag is used, the hitlist will include molecules electrostatically similar to the query.
Before executing the search, the GUI also writes the encapsulated query into the query file.
By default, an information file titled brood_1.info is written or updated for every 500 fragments that are processed. Reading this info file allows a user to monitor the progress of the database search while it is occurring and it serves as a record of the performance of that search after the execution is completed. During execution, the GUI uses this file as well as direct communication with the application to keep the user apprised of all information available in the info file.
An example info file is shown below.
*********** Progress Update *********** Percent of Database Processed = 100% Total packets read = 329 Packets suitable for Processing = 99 Number of Fragments Overlayed = 92134 Number of Fragments Eliminated = 74560 Number of Fragments Processed = 166694 Number in color hitlist = 300 Remove for quick search = 291 Remove protein clashes = 0 Remove select protein = 0 Remove duplicates = 0 Remove unstable bonds = 0 Remove strained molecules = 0 Number in final hitlist = 9 Number of Warnings = 2 Number of Errors = 0 Processed fragments/sec = 1742 Elapsed Time (sec) = 95 ***************************************
If any warnings or errors are noted in the info file, it is strongly suggested that a user check in the log file to determine the nature of the problem!
BROOD‘s command-line interface can be efficiently run using the -param command line parameter followed by the name of a parameter file. Param files are files that contain one command-line parameter on each line. Every execution of BROOD, including executions started within the GUI, generates a .param file called brood.param. This file contains all of the parameters used by BROOD. Further, this file can be used in subsequent runs with the -param flag either with or without user modifications.
The .param file is particularly useful if you want to use the graphical interface to set up a job, but execute the job on another machine (such as a cluster). The .param file can be moved to a different machine along with the -queryMolecule file it specifies and it can be used to execute exactly the same run as would have been executed from within the graphical interface.
The following is an example of a .param file generated by the GUI (brood_1.param). It can be used to regenerate the run using the command-line below.
This file could be edited or used “as-is” with subsequent runs. In addition, any explicit command-line parameter takes precedent of parameters in the .param file, so this mechanism can be used to execute a series of similar jobs.
prompt> brood -param brood_1.param prompt> brood -mpi_np 16 -param brood_1.param -prefix brood_mpi1
Listing of brood.param file contents:
#Interface settings #Execute Options : -param /Users/chemist/data/brood_49.param -input_chunksize 10 -success_chunksize 10 -failure_chunksize 10 #Brood : #Input : #-queryFrag (Not set, no default) -queryMol /Users/chemist/data/brood_49_query.oeb.gz -db /Users/chemist/data/db/chembl17 #-prot (Not set, no default) #-select (Not set, no default) #-noqueryprot false #-cpddb none #-param (Not set, no default) #Output : -prefix brood_49 -dots true #-log (Not set, no default) #-info (Not set, no default) #-report (Not set, no default) -format oeb.gz -csv true -idea true -neutralpH true -tautomer true -hitlistProt true #Control parameters : -quickLook true -ringOnly true -ET false -linkOnly false -sdTag verbose -checkBond true #-maxHit 1000 #-title (Not set, no default) -attachColor false #-attachFrag true #Advanced parameters : #-bondOrder true -attachmentCutoff 0.78 -shapeCutoff 0.6 -attachmentScale 1.5 -checkGeometry true -fromCT false -fileChrg false -interval 5000 -hitinterval 1000 #-maxFrag 0 -rangeSize 6 -rangeOffset 0 -bumpRadius 2.25 #Property Selection : -property false #-maxMolWt 500.0 #-minMolWt 100.0 #-maxlogp 5.0 #-minlogp -1.0 #-maxpsa 150 #-minpsa 60 #-maxRotBond 13 #-minRotBond 0 #-maxHvyAtom 35 #-minHvyAtom 7 #-maxLipinskiDon 8 #-minLipinskiDon 1 #-maxLipinskiAcc 11 #-minLipinskiAcc 2 #Synthetic Properties : #-maxComplexity 1.0 #-minComplexity 0.0 #-maxFreq 100 #-minFreq 1 #Derived Property Selection : #-maxLipinski 1 #-minLipinski 0 #-maxMartin 1.0 #-minMartin 0.2 #-eganEgg true #-veber false #-maxFsp3C 1.0 #-minFsp3C 0.3 #-maxAromFJCt 5 #-minAromFJCt 0
prompt> brood -param brood.param
This would result in exactly the same execution as the one which generated the file above.
The log file contains all of the critical information about the execution in one place. It begins with a copy of the param file and it finishes with the final info file output. In between, it contains all the warnings and errors that might have occurred during execution. The log file gives a user a single place they can check to determine what job was run, if it executed properly and how long it took.
The report file contains a detailed listing of the similarity scores for every molecule in the database that was sufficiently similar to warrant a 3D overlay. By default, the file is titled brood_1.rpt and contains 1 line for the column titles and 1 line for each molecule in the database (tab separated columns). The report file contains columns for
- database fragment SMILES
- database fragment title
- query SMILES
- number of attachment bonds
- structural rms value
- attachment score
- shape Tanimoto
- color Tanimoto
- combo score (shape + color)
- et attachment score
- et shape Tanimoto
- electrostatic Tanimoto
- et combo score (shape + et)
- a comment regarding the disposition of the fragment (if and why it failed to be scored)
If a particular score was not calculated for any given fragment, “-” will be found in the report file under the corresponding column. Since all the data for a fragment is contained in a single line, there is only one report file regardless of the number or type of the number or type of hitlists generated. This file is in tab separated format and can easily be imported into a spreadsheet program for further analysis of the results.
As discussed above, a hitlist is generated for each execution of BROOD. All of the hitlists are written in the file format specified by -format, which defaults to gzipped OEB format .oeb.gz. If the -queryMolecule flag is specified, the hitlist will contain complete molecules, otherwise it will contain fragments. The first entry in the hitlist is the query molecule or query fragment. Each subsequent analog molecule (fragment) in the hitlist is oriented in the optimum overlay on the query molecule (fragment). For the “struc” hitlist, this overlay is the optimum overlay of the attachment point atoms. In addition, by default, the similarity scores and physical property data for each molecule (fragment) are attached as SDTags (OEB and SD format only). In the case when the queryMolecule is not specified, but the -ET flag is set, in the hitlist, the attachment vectors of the fragments are replaced by methyl groups. This facilitates easy calculation of electrostatic potentials in data visualization programs such as VIDA.
By default, the hitlist files are written periodically while the search is being carried out (see -hitinterval). This allows a user to examine results at intermediate stages without waiting for the entire search to complete.
If the -queryMolecule flag specified a 2D input molecule, then all of the analog molecules in the hitlist will likewise be in 2D format, though the fragment similarity search will have been carried out in 3D. When using the graphical interface with a 2D input molecule, a 3D molecule is generated prior to processing and the output format will always be 3D.
At the end of the primary fragment search, a complete hitlist is generated. In post-processing, the fragments are built into whole molecules and tested for strain, the newly formed bonds are examined for chemical stability, potential duplicates in the hitlist are removed, the new molecules are checked for bumps with the protein environment and against the optional selectivity protein. At any of these stages fragments that are similar according to the primary search criteria may be removed from the hitlist for these secondary criteria. In rare cases, few fragments may remain in the final hitlist and the user may want to examine the removed fragments without needing to re-run the entire search. For this reason, any fragment removed at the post-processing stage is written to the removed hitlist brood_1.removed.oeb.gz.