JPH0342034A - Method for predicting chemical structure of organic compound - Google Patents

Abstract

PURPOSE:To facilitate input work and to reduce the calculating time by accumulating the partial structure of an org. compd., position of its absorption spectrum and magnitude of energy to the spectral information in the partial structure estimating part. CONSTITUTION:The sample information A used for prediction is inputted to a computer C. The information A is resolved into an estimated partial structure by the partial structure estimating part 1 in the computer C, and preassembled into a candidate structure in the candidate structure assembling part 2. The inadequate part is then rejected in a structure testing part 3, and all the desired structural isomers are outputted from the output parts 4 of the cathode-ray tube, printer, etc. When the partial structure IA of the material is known, the information is immediately inputted to the candidate structure assembling part 2 without passing through the partial structure estimating part 1.

Description

[Detailed Description of the Invention] [Industrial Application Field] This invention is used in the field of organic chemistry to predict the chemical structure of organic compounds used to predict the structure of naturally occurring substances or substances obtained in synthesis or reaction processes. The present invention relates to a prediction method, and particularly to a chemical structure prediction method suitable for predicting the chemical structure of an organic compound using a computer.

[Conventional technology]

It is generally known that even substances with the same molecular formula do not necessarily have a single structure, but have several types of bonding structures. Recognizing and understanding the chemical structure of unknown substances is a necessary process for obtaining desired substances in the field of organic synthetic chemistry, and is carried out as a daily development activity.

Further, since the analysis work requires a lot of labor, various studies are being conducted on the use of computers to ensure speedy and accurate work. In other words, it has been considered to accumulate data on a large number of substances with known chemical structures, the samples used to produce them, and the reaction processes, and search for samples with the corresponding chemical structure from among them, but there are countless Having the chemical structure of a substance as information would be a waste of memory, no matter how computer-oriented, and it was difficult to expect complete information in terms of chemistry.

Therefore, the lead inventors first prepared an information file that contained data about the chemical reaction products, the partial structures of the samples used to generate them, and the regularity of their bonds, and from that file the samples arrived. He invented and filed a patent application for a chemical structure prediction method for searching and predicting the chemical structure of organic compounds to be obtained (Japanese Patent Application No. 1983-243468).

[Problem to be solved by the invention]

However, input for the sample relies exclusively on known molecular formulas and partial structures, and there is a problem in that sample data cannot be input when the sample is an unknown material.

[Means to solve the problem]

This invention solves the above-mentioned problems and makes it possible to numerically input chemical properties such as known partial structures and molecular formulas, as well as infrared absorption spectra and nuclear magnetic resonance spectra of samples, thereby enriching the input data. The purpose is to combine the partial structure estimator, which searches the information file for the partial structure of an organic compound based on the input material information and estimates it, and assembles the material structure by combining the partial structures obtained there. The partial structure estimation unit outputs a desired substance structure through a candidate structure assembly unit and a structure verification unit that excludes candidate compounds obtained therefrom that do not correspond to the properties of the input substance information. The spectral information stores the partial structure of the organic compound and the position and energy of its absorption spectrum, and the partial structure estimator is configured to be able to input at least either the molecular formula or the spectral information numerically. It is characterized by the fact that

[Effect]

In computer C, each partial structure is digitized based on absorption, signal position, and intensity and stored as information.

When material information A is given as spectrum information 1b,
Only suitable information is selected from the accumulated information and the partial structure is estimated.

Then, in the next structure assembly section, that partial structure is combined with other partial structures according to the priority order from among the accumulated connection information, and inappropriate ones are removed and output.

〔Example〕

The present invention will be explained below with reference to illustrated embodiments. 1st
The figure is a flowchart of this chemical structure prediction method, and sample information A used for prediction is input to computer C. The sample information A is decomposed into estimated substructures in a substructure estimating section 1 in a computer C, tentatively assembled into a candidate structure in a candidate structure assembling section 2, and then inappropriate parts are removed in a structure verification section 3. All desired structural isomers are outputted from an output section 4 such as a cathode ray tube or printer. Note that Ci is an internal memory of the computer C, and constitutes an information file that stores information on partial structures, estimated partial structures of sample information A, etc., which will be described later.

The sample information A includes a partial structural formula 1a, spectral information 1b such as an infrared absorption spectrum or a nuclear magnetic resonance spectrum,
The molecular formula is 1c, etc. When the partial structural formula 1a of a substance is known, it is immediately inputted to the candidate structure assembly section 2 without going through the partial structure estimating section 1, but when it is based on the spectrum information 1b or the molecular formula lc, the candidate structure Prior to the process to the assembly section 2, the partial structure is analyzed in a partial structure estimating section 1, and a partial structural formula 1a is estimated.

Here, the partial structural formula 1a of the sample information A is the structure of the main part that is sufficient to assemble the structure of an almost infinite number of organic compounds, and the partial structure estimation unit 1 calculates the structure from the sample information A through the process shown below. It is broken down and organized into formulas. In this example, the structural formula consists of three types of hierarchies. The first layer is the basic partial structure, and as shown in Figure 2, it consists of C, H, O, N, S,
Fifteen types of substructures consisting of halogens, etc. were selected empirically, and the second layer was created based on the attributes when searching for their bonding partners, that is, the properties of the bonds of atoms with free bonding hands in each substructure. As shown in Figure 3, 86 types were selected, and in the third layer, the partial structures of the second layer with free bonds are selected according to the priority order of selection as a bonding partner, taking into account the attributes of the bonding of the partner piece. total of 6 as shown in Figure 4.
Thirty types of partial structures are set.

The attributes for finding a binding partner are as shown in Figure 5.
Iodine, nitrogen, oxygen, aromatic carbon, carbonyl carbon, s
Twelve species ranging from I to CS, such as p 2 carbon, sp carbon, and sp 3 carbon, are listed in order of priority.

These 630 types of partial structures cover all conditions such as being able to estimate their existence from spectral data, having no overlapping parts, and being able to cover the target chemical substances by combining them. FIG. 6 shows an example of a property table that organizes the partial structural formulas. For each substructure, the properties of the bond itself and the properties required of the bond partner are arranged in correspondence.

In the present invention, as the sample information A input to the computer C, in particular, spectrum information 1b such as an infrared absorption spectrum (IR) or a nuclear magnetic resonance spectrum (NMR) can be input. According to the spectrum of isophorone (C9H140) shown in Figure 7, the infrared absorption spectrum (IR) shown at the left end has the position (am) of the downward protruding valley and the absorption intensity (%T) as shown in the lower column. will be converted into data. Similarly, in hydrogen nuclear magnetic resonance spectrum (HNMR), the position of the peak and its area intensity are converted into data, and in addition, in (BCNMR), the multiplicity of the peak is converted into data in addition to the position of the peak and its area intensity.

Next, the operation of the method for predicting the chemical structure of organic compounds using computer C will be explained. When the sample information A is input to the computer C, the partial structure estimating unit 1 calculates the molecular formula 1c and spectrum information 1b of the input sample information A from among the 630 types of partial structures listed in the characteristic table of the partial structural formulas. The substructure is estimated by an elimination method that removes contradictions. Among the provided sample information A, partial structures that do not fall within predetermined tolerances based on the numerical data stored in the internal memory Ci, such as the position and intensity of absorption, are erased, and the remaining structures are estimated. It is output to the next process as a partial structure.

When input is made using the molecular formula 1c, among the chemical structures accumulated in the information file Ci of the partial structural formula, partial structures that do not match the molecular formula are deleted, the desired partial structure is estimated, and the next candidate structure is generated. It is output to the assembly section 2.

In addition, apart from the above routine, if a partial structure that cannot exist in the sample information A or a partial structure that always exists is known, the partial structure can be inputted into the information file (internal memory) C1. Of course, the accuracy of estimation is improved.

In the candidate structure assembling unit 2, partial structures having mutually determined bonds from the data of the second layer and the third layer are combined in accordance with the bonding priority order and output as structural isomer candidates.

That is, the property shown in the middle column of FIG. 6 is simply compared with the property of the binding partner shown in the rightmost column, and the matching properties are combined and output according to the priority order of combination.

In addition, the structure verification section 3 calculates the number of carbons, the number of saturated carbons and hydrogens, the number of oxygens, etc. necessary for the existence of spectral information and partial structure among the candidate assembled structures outputted from the candidate structure assembly section 2. conditions, and furthermore, when partial structures that cannot exist in sample information A or partial structures that always exist are known, those that cannot exist are removed according to that information, and only correct isomeric structures are output to the output section 4. is output from.

〔Effect of the invention〕

As described above, according to the present invention, the absorption spectrum of each 4° partial structure of an organic compound is quantified by the position and energy magnitude, so unique data for each partial structure is stored in an information file (internal memory ) If stored in C1, the estimated partial structure can be quickly retrieved from the manually input spectrum information 1b. Therefore, in addition to known partial structures and molecular formulas, infrared absorption spectra and nuclear magnetic resonance spectra of the sample can be used to estimate the partial structure, making it possible to enrich the input data. In addition, since the absorption position and its energy are used as the information source as the input method using the spectrum information, the input work is easy, the storage capacity of the computer is small, and the calculation can be completed in a short time. There are effects such as being able to.

[Brief explanation of drawings]

The drawings show an embodiment of the present invention, and FIG. 1 is a flowchart of the prediction method according to the invention, FIGS. 2, 3, and 4 show partial structure tables, and FIG. 5 shows a partial structure table. FIG. 6 is a characteristic table showing the attributes of the structure, FIG. 6 is an arrangement table for organizing partial structures, and FIG. 7 is a characteristic chart showing spectrum information and its arrangement table.

Claims (3)

[Claims] (1) A partial structure estimating unit that searches and estimates a partial structure of an organic compound based on input substance information from an information file; a candidate structure assembly unit that assembles a substance structure by combining the partial structures obtained there; and Among the candidate compounds obtained therein, the structure verification section eliminates those that do not correspond to the characteristics of the input material information, and the desired substance structure is output. The partial structure of the organic compound and the position and energy size of its absorption spectrum are accumulated, and
The partial structure estimation unit is configured to allow input of at least a molecular formula or spectral information numerically. (2) In the information file used in the structure assembly section, parameters indicating the properties of the substructure of the substance and parameters indicating the properties of the other party to which the substructure can be bonded are stored, and based on these, 2. The method for predicting the chemical structure of an organic compound according to claim 1, wherein the partial structures inputted from the partial structure estimating section are combined to output a candidate substance structure. (3) The structure verification unit is configured to remove candidate substance structures having properties that do not match the substance information numerically input to the partial structure estimation unit. A method for predicting the chemical structure of organic compounds.