OPERA v2.8.4

(See the install and quick run guide pdf file in the zip file for more info and input options)

Clarifications about log4j concerns:

1. The presence of a log4j jar file on a computer does not imply a vulnerability in itself. It’s a very common file in java-based tools. It is only when log4j is used on an exposed server that the vulnerability can be a problem.
2. We do not use log4j in OPERA software. OPERA runs locally and does not connect to the internet. Our testing thus far indicates that the removal of the log4j.jar file will not affect OPERA software. OPERA should work normally with or without the log4j file as it does not depend on it.
3. OPERA uses two main tools: KNIME and MATLAB. In OPERA 2.8.4, KNIME was updated to version 4.5.1 which updated the log4j file to deal with the vulnerability in the previous versions. For more details see https://www.knime.com/changelog-v45. For the MATLAB runtime, MathWorks has published the following in the Trust Center (version 3 of 2021-12-18): https://www.mathworks.com/content/dam/mathworks/policies/mathworks-response-to-cve-2021-44228-log4j-vulnerability.pdf

To scan and remove any unwanted files/classes you can use: https://github.com/logpresso/CVE-2021-44228-Scanner

Installer files to choose from:

• OPERA2.8_UI(_win.zip for Windows, .tar.gz for Linux) (recommended for most users): Graphical User Interface installer. The installer will automatically install all dependencies (it is self-contained with mcr runtime, if mcr runtime v99 is not already installed, it will automatically install it). No internet is required during installation, suitable for firewalled/unconnected machines. Once installed, the models run LOCALLY and the app has NO access to the web.

• OPERA2.8_CL(_win.zip for Windows, .tar.gz for Linux): Command line installer. The installer will automatically install all dependencies (it is self-contained with mcr runtime, if mcr runtime v99 is not already installed, it will automatically install it). No internet required during installation, suitable for firewalled/unconnected machines. Once installed, the models run LOCALLY and the app has NO access to the web.

• OPERA2.8_CL_Par(_win.zip for Windows, .tar.gz for Linux): Command line installer for the parallel computing version to maximize speed by using multiple cores in parallel (recommended only for files with a large number of chemicals). The installer will automatically install all dependencies (it is self-contained with mcr runtime, if mcr runtime v99 is not already installed, it will automatically install it). No internet is required during installation, suitable for firewalled/unconnected machines. Once installed, the models run LOCALLY and the app has NO access to the web.

• libOPERA2.8_*.tar.gz: additional libraries in C, C++, Java, and Python to embed in other projects.

Updates:

v2.8

• New Caco-2 permeability (logPapp) model
• Updated human plasma fraction unbound (Fub) model with new data
• Updated human hepatic intrinsic clearance (Clint) model with new data
• Updated the Knime version in the QSAR-ready standardization workflow to 4.5 to fix the log4j java vulnerability
• Fixed bugs and exceptions.

v2.7

• New Fub and Clint models based on newly curated data
• CERAPP, CoMPARA, and CATMoS: confidence index updated to include experimental data information about the 5NN.
• CATMoS: confidence interval updated to 0.25 (log10 mg/kg) based on new variability analysis study (Karmaus et al 2021, in prep)
• All other models (except CATMoS and categorical) include a confidence range (based on weighted st-dev of 5NN).
• Improved speed
• Updated MATLAB mcr v99, more compatibility with OS updates
• Better packaging of C/C++, JAVA and Python libraries to support output results in objects (in addition to files)
• Fixed bugs and treated new exceptions

v2.6

• Improved clarity of error messages.
• Implementation of the latest QSAR-ready structures standardization workflow for all versions (CL, parallel CL, and GUI in Linux and Windows). For more info about the workflow: https://github.com/kmansouri/QSAR-ready.

v2.5

• For all models: used the latest QSAR-ready structures using the updated (v2) standardization workflow (https://github.com/kmansouri/QSAR-ready).
• Added the latest DSSTox update including ~900k structures to be searched using the input option as text file (.txt) with molecule identifiers (CASRN, DTXSID, DTXCID, or InChIkey).
• When the search by ID is used, a Found-by column is added to the output for additional information about the found and not found or inadequate structures (inorganics, mixtures...)
• Added new data (for PFAS and other structures) for LogP, water-solubility, vapor pressure, and melting point.
• Improved detection of ionization sites for pKa prediction
• Fixed a bug in LogBCF model on Windows GUI that was preventing from writing out the results.
• Trim IDs (CAS or DTXSIDs) to maximize matches to retrieve QSAR-ready structures and/or experimental data
• Optimized loading time and calculation speed

v2.4

• Improved detection of ionizable groups for pKa prediction
• Trim IDs (CAS or DTXSIDs) to maximize matches to retrieve QSAR-ready structures and/or experimental data
• Added test to warn about exceeding the recommended molecule size limit by CDK to avoid failed descriptor calculation or long processing.
• Updated CATMoS with a Weight-of-Evidence approach (WoE) to combine predictions from all 5 endpoints
• Optimized loading time, calculation speed, and handling of chemical descriptors

v2.3

• Retrieve salts information (if not provided) for MP predictions based on structure ID in the input file
• Fixed GUI bugs and added tests for molecular descriptor limitations.
• Added output option to get Experimental values if CASRN or DTXSID is provided in the input
• Added input option as text file (.txt) with molecule identifiers (CASRN, DTXSID, DTXCID, or InChIkey) available for >800k structures

v2.2

• Treated exceptions when PaDEL descriptors fail on certain structures (inorganic, mixtures...)
• Flexibility to change the install directory
• Added new models (FuB and Clint)
• Fixed GUI bugs
• Improved memory handling (files with >10,000 molecules)
• Added a correction patch for PFAS chemicals for logP and WS models
• Corrected CATMoS-LD50 experimental values
• Fixed bug related to running all models as default

v2.1

• Added a graphical user interface
• Added Clint and FUB models

v2.0

• Added pKa, logD, CERAPP, CoMPARA, CATMoS

v1.5

• Added RT model
• Updated physchem properties models: logP, logBCF

v1.0

• Physchem properties

Tool requirements:

• System: Windows/Linux (64bit) not tested on old versions (prior to 2013).
• Java JRE/JDK 1.5 or higher (64bit).
• Input QSAR-ready structures (can be downloaded from https://comptox.epa.gov/dashboard): * Smiles file (tab-delimited text file with no headers and no empty lines, .smi extension) * SDF file (text file with 2D or 3D atom coordinates, .sdf extension) * Text file with molecule identifiers: CASRN, DTXSID, DTXCID, or InChIkey (one column with no headers and no empty lines, .txt extension).

Limitations of the current version:

• Molecular descriptors can only be calculated for: * Maximum molecular weight 2000 g/Mol * Maximum number of heavy atoms <= 100 * No nano-materials like shapes (spheres, tubes...)

Models:

*  Latest version OPERA v2.5:

      + Molecular descriptors:  
- PaDEL (2.21) (https://doi.org/10.1002/jcc.21707 )
- CDK (2.0) (https://doi.org/10.1186/s13321-017-0220-4)


      + List of newest models:

- Caco-2 permeability (logPapp) model

- FuB: Plasma fraction unbound (human, fraction)

- Clint: hepatic intrinsic clearance (human, ul/min/10^6 cells)

- pKa: acid dissociation constant

- LogD: Octanol-water distribution constant. LogD is equivalent to logP for non-ionizable compounds.

- CERAPP: Collaborative Estrogen Receptor Activity Prediction Project. Binding, Agonist, and Antagonist Estrogen Receptor activity (https://ehp.niehs.nih.gov/15-10267/)

- CoMPARA: Collaborative Modeling Project for Androgen Receptor Activity. Binding, Agonist, and Antagonist Androgen Receptor activity (https://doi.org/10.13140/RG.2.2.19612.80009, https://doi.org/10.13140/RG.2.2.21850.03520)

- CATMoS: Collaborative Acute Toxicity Modeling Suite. Very-Toxic, Non-Toxic, EPA categories, GHS categories, LD50 (Log mg/kg) (https://doi.org/10.1016/j.comtox.2018.08.002)

- Structural Properties: MolWeight, nbAtoms, nbHeavyAtoms, nbC, nbO, nbN, nbAromAtom, nbRing, nbHeteroRing, Sp3Sp2HybRatio, nbRotBd, nbHBdAcc, ndHBdDon, nbLipinskiFailures, TopoPolSurfAir, MolarRefract, CombDipolPolarizability.


      + Previous models (since v1.5):


- OH (LogOH) in cm3/molecule-sec: The OH rate constant for the atmospheric, gas-phase reaction between photochemically produced hydroxyl radicals and organic chemicals.

- BCF (Log): Fish bioconcentration factor

- Biodeg (LogHalfLife) in days: biodegradation half-life for compounds containing only carbon and hydrogen (i.e. hydrocarbons). 

- Ready_biodeg (classification: 0/1): Ready biodegradability of organic chemicals. 

- BP in deg C: Boiling Point at 760 mm Hg

- HL (LogHL) in atm-m3/mole: The Henry’s Law constant (air/water partition coefficient) at 25C

- Km (Log KmHL) half-lives in days: The whole body primary biotransformation rate constant for organic chemicals in fish. 

- KOA (Log): The octanol/air partition coefficient.

- LogP (Log): Octanol-water partition coefficient, log KOW, of chemicals.

- MP in deg C: Melting Point

- Koc (Log) in L/Kg: the soil adsorption coefficient of organic compounds.  The ratio of the amount of chemical adsorbed per unit weight of organic carbon in the soil or sediment to the concentration of the chemical in solution at equilibrium.

- VP (Log) in mmHg: Vapor Pressure experimental values between 15 and 30 deg C (majority at 25-20C)

- WS (Log) in Molar moles/L: Water solubility at 25C. 

- RT in minutes: HPLC retention time.