Contact us
The software uses machine learning to help users solve problems in organic chemistry
Functional Purpose of the Software
The software assists users with the following processes:
— processing and analysing of chemical information using machine learning models;
— experimental modeling of chemical compounds with specific properties.
Syntelly provides information on existing chemical compounds, scientific literature, and chemical reactions for searching.
The Syntelly software suite is intended for use in research centers, higher education institutions, and enterprises of the chemical, cosmetic, and pharmaceutical industries for finding known chemical information and modeling promising candidate molecules for new substances and materials (ncluding the development of novel pharmaceutical products).
The platform can also be used by organizations engaged in intellectual property protection and patent search, as well as by governmental, commercial, and educational organizations.
Operational Purpose of the Software
9
Chemical Space Visualization Module "SynMap"
01
This is a visual module for the analysis of chemical space (2D/3D), based on a pre-trained neural network model. It enables the main groups of chemical compounds present in a dataset to be represented quickly and intuitively.
02
The model projects chemical structures onto an X-Y coordinate system in two dimensions. The tool allows molecular datasets to be compared and analysed by overlaying them on the map as different layers.
02
Includes a generator of new structures with predefined parameters (QED, CATS, boiling point, melting point, mouse oral LD50, LogP, LogS, DMSO solubility, complexity, SYBA and BRUTTO)
8
Datasets Module
01
Supports batch data processing and functionality for working with datasets of molecules and chemical reactions.
02
Import formats: SDF, CSV, SMI.
Export formats: SDF, CSV, XLSX.
Export formats: SDF, CSV, XLSX.
03
Supports three types of datasets:
- thematic datasets — grouped sets of structures by known biological targets;
- company datasets — datasets for collaborative work within one organization, including activity logging and role-based access control;
- personal datasets — datasets for individual use with the ability to add personal experimental data to molecules.
It provides tabular analysis of structures by all properties, with filtering and colour-coding based on specified conditions, and subsequent export of the filtered dataset.
7
Molecular Editor Module
- input and editing of molecular structures using a graphical interface;
- calculation of all predicted properties for structures not present in the Sintelly database (new user-defined structures).
6
Reaction Search Module
01
Use the molecular editor or identifiers (SMILES, IUPAC name, CAS number, InChI, etc.) to search for chemical reactions by reactant or product structures.
02
Search can be configured by the role of the substance in the reaction and by reaction yield percentage.
03
The module provides information on each reaction from the literature, including reaction conditions (temperature, pressure, catalyst, etc.), reagents and solvents used, as well as access to relevant scientific literature and experimental protocols.
04
The search history can be saved with the input type fixed.
5
Literature Search Module
01
The literature database includes publications, patents, and patent applications:
- 160 million publications;
- 20 million patents.
02
Search parameters include:
- document type (article/patent);
- publication language (Russian/English);
- publication date;
- fields (title, abstract, author, DOI, patent number, journal, publisher, owner, applicant, full text, formula).
03
The module provides full-text search functionality for patent documentation.
04
Users can view the source describing the searched compound (e.g., structural formula, literature linkage).
05
An advanced query builder enables creation of complex combined queries (e.g., structure + keyword) using logical operators (AND / OR / NOT).
06
The search history can be saved with the input type fixed.
4
Structure Search Module
01
The module is based on a property database of already studied compounds and provides fast access to compound data, including:
- structures of organic compounds (over 160 million records);
- experimental data (over 2.4 million records).
02
Available filters:
- exact match;
- substructure search;
- similarity search;
- Markush structures.
03
Search can be performed using a molecular editor or a search string in any convenient format: keywords, synonyms, IUPAC name, CAS number, SMILES, InChI, etc.
04
The search history can be saved, with the input type (structural formula, SMILES, etc.) fixed.
3
Reaction Card
Contains the following information:
- reaction scheme;
- yield;
- sources;
- reaction conditions (temperature, pressure, catalyst, reagent, solvent, pH, equipment, atmosphere, irradiation, time).
2
Literature Card
Includes the following information:
- title of the publication or patent;
- abstract;
- structures from the article;
- source;
- publication date;
- patent number or article DOI.
Some patents are also available in full text.
1
Structure Card
01
Each structure card provides the following information:
- structural data block — all available structure identifiers (CAS number, IUPAC name, SMILES, InChI, InChIKey, molecular formula, molecular weight);
- known synonyms under which the structure may be listed in other databases;
- links to external databases;
- 10 physicochemical properties;
- 5 environmental properties;
- 40 toxicity indicators;
- 18 biological properties (including 5 cytochromes);
- two different methods for assessing synthetic complexity;
- drug-likeness parameters (Lipinski’s Rule of Five, Ghose filters, Oprea rules, Veber rule, PAINS).
02
The following information is available on each structure's card:
In total, more than 80 properties are presented in each structure card. If experimental values for the queried molecule are available in the database, the system displays them with a green indicator next to the relevant parameters.
03
If no experimental data is available, predicted values calculated by deep neural networks are displayed.
04
The model applicability indicator compares each specific structure with the training dataset and shows the applicability of the model to the selected structure as a percentage.
05
All presented information can be downloaded as a PDF report.
№
Product Component
Functional characteristics
Functional Characteristics of Software Components
13
PDF2SMILES (OCR) Module
01
A tool using Optical Character Recognition (OCR) for extracting structural formulas from PDF documents.
02
The module provides:
- upload of documents up to 100 MB;
- reliability assessment for each recognized structure;
- editing of recognized structures;
- storage of recognized user documents (document collections);
- download of recognition results in PNG and CSV formats;
- saving recognized structures into a separate dataset.
14
SMILES to IUPAC Module
Generates systematic IUPAC nomenclature according to IUPAC rules in English.
15
Statistics Module
Provides statistical parameters (metrics) for machine learning models, including RMSE and ROC-AUC, displayed in the structure card.
16
Structure Comparison Module
Displays statistical parameters (metrics) of machine learning models (RMSE, ROC-AUC) and allows selection of specific characteristics for comparison from the list of properties.
12
Synthesis Cost Module
01
This tool estimates the cost of chemical synthesis. Users specify synthesis parameters such as the product, reagents, desired product weight and number of reaction stages.
02
The result is a list of the top five reaction schemes, ranked by cost, which enables the most economically efficient synthesis route to be selected.
02
The module provides a detailed analysis of each scheme and allows users to edit the cost table and export data in Excel, PDF and CSV formats.
11
Spectra Module
Predicts spectral data, including:
01
NMR (¹H, ¹³C, ¹⁵N, ¹⁹F) for small organic molecules. Results are presented as “chemical shift – relative intensity”; multiplicity is also predicted for ¹H spectra.
02
Mass spectrometry with configurable parameters: spectral type, ion mode, adduct type, RI rounding, and m/z.
03
Infrared spectrometry with various acquisition methods (gas, liquid, CCl₄, KBr).
04
All presented information can be downloaded in PDF format.
10
Reaction Prediction Module
01
Provides planning of organic compound synthesis using a neural network model, including two options:
- Synthesis — prediction of potential reaction products based on the reacting reagents (single-step organic synthesis);
- Retrosynthesis — analysis of synthesis routes for the target molecule, including reaction trees and verification of commercial availability of reagents.
02
All information presented can be downloaded in PDF format.