CN110379467B - Chemical molecular formula segmentation method - Google Patents

Abstract

The invention discloses a chemical molecular formula segmentation method, which is implemented on the basis of a molecular formula graph structure, namely, a certain group of a molecular formula is cut off by using a specific type of chemical reaction, the segmentation simulates a sub-process in the chemical reaction, and in the real chemical reaction, a compound which has undergone the chemical reaction is cut off partially and then is combined with other compounds to form a new compound. The invention realizes the segmentation of molecular formula on the graph structure of atoms and chemical bonds, has strong operation flexibility and strong customizability, and is suitable for the segmentation of various situations. And finding out which branches on which atoms in the chemical formula to be split are to be split and which branches are to be reserved according to the reaction type string representing the reaction type in the chemical formula to be split and the atom mapping relation between the reaction type strings representing the conversion type in the chemical formula to be split, so as to realize correct splitting.

Description

A kind of chemical molecular formula segmentation method

technical field

The invention belongs to the technical field of compound synthesis, and in particular relates to a chemical molecular formula segmentation method.

Background technique

In the process of compound synthesis, the reagent molecules with protective groups should be deprotected and reacted with each other to obtain new compounds. From a chemical point of view, both steps are chemical reactions; from a computational point of view, the chemical reaction in which a molecule removes protecting groups is a "site cleavage" of the molecule, resulting in a molecule that can react (link) with other molecules. site, and then react with each other is "site splicing" between molecules, linking sites where chemical reactions can occur to obtain new compounds.

The current processing method is to record the molecules with sites obtained in the first step, and mark the sites obtained by different reactions. For example, the site obtained by reaction type 1 is marked as [R1], and the position of the site is Molecules remove part of the atomic positions in the reaction, and then splicing molecules with the same site label, such as two molecules with [R1] can be spliced.

The current computational chemistry tools, such as Openbabel, RdKit, etc., are segmented by simulating the real chemical reaction process, that is, inputting the compound to be reacted, and obtaining the reacted compound.

SUMMARY OF THE INVENTION

The purpose of the present invention is to aim at the deficiencies of the prior art, and realize a segmentation method on the basis of the graph structure of the molecular formula, that is to use a specific type of chemical reaction to cut off a certain group of the molecular formula, and the segmentation simulates the chemical reaction. A sub-process, in a real chemical reaction, a part of the chemically reacted compound is cut off and then combined with other compounds to form new compounds. In the application scenario of compound synthesis through a large number of reagent molecules, molecular formula segmentation is an inherent sub-link. Removing a fixed group from a reagent molecule must first remove a certain group before it can be combined with other compounds.

The object of the present invention is achieved through the following technical solutions: a chemical molecular formula segmentation method, for a molecular formula smi with or without a site, the site in the molecular formula is recorded as [Ri], i=0 ,1…N, [Ri] corresponds to a certain reaction type, and N is the total number of reaction types;

For a reaction type Reaction, define a reaction type string smi_sub and a conversion type string smi_convert, indicating that if there is a smi_sub substring in smi, then convert the substring part to smi_convert; the smi_sub substring contains at least one connection with the rest of smi The site [Ri] of , smi_convert has the same site [Ri] in the corresponding position, and smi_convert also contains the atom [*] to be excised. The segmentation process is as follows:

(1) Read in smi and smi_sub, convert them into graph structures mol and mol_sub, and generate the atomic mapping table maplist2 from mol_sub to mol. If there is no mol_sub substructure in mol, no further segmentation will be performed. If there are multiple mol_sub substructures in mol, multiple mapping relationships are saved in maplist2. When mapping, consider whether the atom corresponding to the linking site satisfies the restriction of the [Ri] type in smi_sub. If not, the mapping fails. When reading in smi_sub, save the position of the ligation site [Ri].

(2) Read in smi_convert, convert it to the graph structure mol_convert, generate the atom mapping table maplist1 from mol_convert to mol_sub, find the position cut_idx of the cut atom [*], and find the position cut_connect_idx of the atom connected by the cut atom [*]. For each atom in mol_convert, when no corresponding atom is found in mol_sub, use -1 to represent the mapped atomic position of that atom.

(3) For each mapping relationship in maplist2, a segmentation result is obtained:

(3.1) Add mol and mol_sub to get the updated mol.

(3.2) Determine whether the mapping relationship will cause the ring structure in the mol to be cut. If so, skip this mapping relationship and execute the next mapping relationship, otherwise, go to step 3.3.

(3.3) Through this mapping relationship, collect and map the positions of atoms that are not cut in mol, that is, the positions corresponding to elements >-1 in maplist1, and store them in not_cut_idxs.

Collect and map the position of the atom connected to the cut atom [*] in the mol, that is, the atom corresponding to cut_connect_idx, and store it in cut_connect_idxs.

Collect the positions of all the cut atoms mapped to the mol, that is, the positions corresponding to the elements of -1 in maplist1, and store them in cut_idxs. For the atoms at the cut point position, store the atoms in the adjacent atoms that are not in not_cut_idxs in cut_idxs. Each cut position maintains a list of cut_idxs, which is empty if the corresponding atom in the mol cannot be found at that position.

For the cut_idxs of each excision position, if it is empty, the hydrogen atom at the position is cut off. If there is no hydrogen atom at the position, the division of this mapping relationship is skipped and the next mapping relationship is executed. If not empty, collect atoms to be cut at this position, including atoms at this position and atoms adjacent to it but not in cut_connect_idxs, recursively collect adjacent atoms of adjacent atoms at this position.

Collect the positions of the atoms to be cut in the mol, that is, those atoms in cut_connect_idx, including atoms at this position and atoms adjacent to them but not in cut_connect_idx and cut_idx, these atoms have new positions in the new mol, recursively collect the atoms The adjacent atoms of the adjacent atoms in the position.

(3.4) Connecting atoms: Connect the corresponding atoms in cut_connect_idxs and cut_idx with chemical bonds.

(3.5) Copy bond properties: Copy bond properties from mol_sub to the newly added bonds in mol. Collect newly added atoms, that is, atoms at both ends of the newly added bond.

(3.6) Update the \ and / steric properties of the double bond: If the steric property structure of the mol contains the atoms to be excised, the steric properties need to be updated. The double bond space attribute structure stores the atoms at both ends of the double bond and the atoms connected to them. If there is a bond between the two ends of the double bond space attribute structure and the newly added atom in the previous step, the corresponding atoms in the structure are stored. Replace with new atom.

(3.7) The collected atoms to be excised are excised.

(3.8) Output a segmented molecule of the maplist2 mapping relationship.

Further, according to the characteristics of each reaction type, make some restrictions on the atom type represented by the site [Ri] in the reaction type string smi_sub, that is, make some restrictions on the type of atom in the molecular formula to be cut corresponding to [Ri], Including the site [Ri] cannot be hydrogen atom (H), or can only be C, N, O atom, or can be any atom, which is limited according to the SMARTS specification.

Further, in the step (1), the read-in molecular formula format is Canonical SMILES, and the input of other formats is converted into this format.

Further, in the step (1), when mapping, consider whether the atom corresponding to the connection site satisfies the definition of the [Ri] type in smi_sub, if not, the mapping fails.

Further, in the step (3.2), it is determined whether the mapping relationship will cause the ring structure in the mol to be cut, that is, to determine whether the mol_sub is mapped to a part of the ring structure of the mol, specifically: through the mapping relationship in maplist2 Find the corresponding atom A of the site to be excised [Ri] in mol. If A has an adjacent atom A_neighbor that is not in the list to be excised, and A_neighbor is on the ring, the ring will be excised.

Further, in the step (3.5), copy the bond properties from mol_sub to the newly added bond in mol, that is, find the atoms at both ends of each bond in mol_sub, find the two corresponding atoms in mol, and combine these two The bond properties between atoms are updated to the property values of the corresponding bonds in mol_sub.

The beneficial effects of the present invention are as follows: the present invention realizes the segmentation of molecular formulas on the graph structure of atoms and chemical bonds, has strong operational flexibility, and is highly customizable, and is suitable for segmentation in various situations. Through the atomic mapping relationship between the chemical formula to be split and the reaction type string representing the reaction type, and the reaction type string representing the conversion type in the chemical formula to be split, find out which branches on which atoms in the chemical formula to be split need to be removed and which branches Branches should be preserved to achieve correct segmentation.

Description of drawings

Fig. 1 is the flow chart of the method of the present invention;

FIG. 2 is a segmentation example of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

A kind of chemical molecular formula segmentation method provided by the invention is as follows:

For a molecular formula smi with or without a site (assuming the input format is Canonical SMILES, input in other formats can also be converted to this format), record the site in the formula as [Ri], i=0,1 ...N, [Ri] corresponds to a certain reaction type, which is obtained by using reagent molecules to carry out this type of chemical reaction, and N is the total number of reaction types;

For a reaction type Reaction, define a reaction type string smi_sub and a conversion type string smi_convert, indicating that if there is a smi_sub substring in smi, then convert the substring part to smi_convert. The smi_sub substring contains at least one site [Ri] connected to the rest of smi, smi_convert has the same site [Ri] at the corresponding position, and smi_convert also contains the atom to be excised [*]. In particular, according to the characteristics of each reaction type, make some restrictions on the atom type represented by the site [Ri] in the reaction type string smi_sub, that is, make some restrictions on the type of atom in the molecular formula to be cut corresponding to [Ri], For example, the site [Ri] cannot be a hydrogen atom (H), or can only be a C, N, O atom, or can be any atom, and the specific limiting treatment is mainly in accordance with the SMARTS specification. The segmentation process is as follows:

1. Read in smi and smi_sub, convert them into graph structures mol and mol_sub, and generate the atomic mapping table maplist2 from mol_sub to mol. If there is no mol_sub substructure in mol, no further segmentation will be performed, indicating that the molecular formula smi cannot have a reaction type. The chemical reaction of Reaction. If there are multiple mol_sub substructures in mol, multiple mapping relationships are stored in maplist2, which means that there are multiple groups in the molecular formula smi that can undergo this type of chemical reaction. When mapping, consider whether the atom corresponding to the linking site satisfies the restriction of the [Ri] type in smi_sub. If not, the mapping fails.

When reading in smi_sub, save the position of the ligation site [Ri].

2. Read in smi_convert, convert it to the graph structure mol_convert, generate the atom mapping table maplist1 from mol_convert to mol_sub, find the position cut_idx of the cut atom [*], and find the position cut_connect_idx of the atom connected by the cut atom [*]. For each atom in mol_convert, when the corresponding atom cannot be found in mol_sub due to being cut off or replaced by a longer substring in maplist1, use -1 to represent the mapped atomic position of the atom.

3. For each mapping relationship in maplist2, get a segmentation result:

3.1. Add mol and mol_sub, that is, stack the data structures of the two to get the updated mol.

3.2. Determine whether the mapping relationship will cause the ring structure in the mol to be cut. If so, skip the mapping relationship and execute the next mapping relationship. Otherwise, go to step 3.3; determine whether the mapping relationship will cause the mol to be cut. The ring structure of mol is cut to determine whether mol_sub is mapped to a part of the ring structure of mol, specifically: find the corresponding atom A of the site to be cut [Ri] in mol through the mapping relationship in maplist2, if A has a phase If the neighboring atom A_neighbor is not in the list to be cut, and A_neighbor is on the ring, the ring will be cut.

3.3. Through this mapping relationship, collect and map the positions of atoms that are not cut in mol, that is, the positions corresponding to elements >-1 in maplist1, and store them in not_cut_idxs.

Collect and map the position of the atom connected to the cut atom [*] in the mol, that is, the atom corresponding to cut_connect_idx, and store it in cut_connect_idxs.

Collect the positions of all the cut atoms mapped to the mol, that is, the positions corresponding to the elements of -1 in maplist1, and store them in cut_idxs. For the atoms at the cut point position, store the atoms in the adjacent atoms that are not in not_cut_idxs in cut_idxs, Since they no longer exist in the transition string, these atoms are also cut off. Each cut position maintains a list of cut_idxs, which is empty if the corresponding atom in the mol cannot be found at that position.

For the cut_idxs of each excision position, if it is empty, the hydrogen atom at the position is cut off. If there is no hydrogen atom at the position, the division of this mapping relationship is skipped and the next mapping relationship is executed. If not empty, collect atoms to be cut at this position, including atoms at this position and atoms adjacent to it but not in cut_connect_idxs, recursively collect adjacent atoms of adjacent atoms at this position.

Collect the positions of the atoms to be cut in the mol, that is, those atoms in cut_connect_idx, including atoms at this position and atoms adjacent to them but not in cut_connect_idx and cut_idx, these atoms have new positions in the new mol, recursively collect the atoms The adjacent atoms of the adjacent atoms in the position.

3.4. Connecting atoms: Connect the corresponding atoms in cut_connect_idxs and cut_idx with chemical bonds.

3.5. Copy bond attributes: Copy the bond attributes from mol_sub to the newly added bonds in mol, that is, find the atoms at both ends of each bond in mol_sub, find the corresponding two atoms in mol, and convert the bonds between these two atoms. The attributes are updated to the attribute values of the corresponding keys in mol_sub. Collect newly added atoms, that is, atoms at both ends of the newly added bond.

3.6. Update the \ and / space attributes of the double bond: Since the space attributes are stored separately and need to be processed separately, if the space attribute structure of the mol contains the atoms to be excised, the space attributes need to be updated. The double bond space attribute structure stores the atoms at both ends of the double bond and the atoms connected to them. If there is a bond between the two ends of the double bond space attribute structure and the newly added atom in the previous step, the corresponding atoms in the structure are stored. Replace with new atom.

3.7. The collected atoms to be excised are excised.

3.8. Output a segmented molecule of the maplist2 mapping relationship, that is, convert mol to CanonicalSMILES format.

An example of segmentation is as follows: As shown in Figure 2, (a), (b), (c), (d) are in turn: reaction type, conversion type, molecular formula to be segmented, and final result of segmentation, where Xx is [ *] atom, S is [R] atom; the corresponding chemical formulas are:

[S]C=C

[S]C([*])C[*]

CCOC(=O)/C=C/C(O)=O

CCOC(=O)C(C(C(=O)O)[*])[*]

The atomic mapping relationship table maplist1 of mol_convert (2nd chemical formula) to mol_sub (1st chemical formula) is {0,1,-1,2,-1}.

It should be stated that the content and specific embodiments of the present invention are intended to prove the practical application of the technical solutions provided by the present invention, and should not be construed as limiting the protection scope of the present invention. Any modifications and changes made to the present invention within the spirit of the present invention and the protection scope of the claims fall into the protection scope of the present invention.

Claims (6)

1. A chemical molecular formula splitting method is characterized in that for a molecular formula smi with or without sites, the sites in the molecular formula are marked as [ Ri ], i is 0,1 … N, [ Ri ] corresponds to a certain reaction type, and N is the total number of the reaction types;

for a Reaction type Reaction, defining a Reaction type string smi _ sub and a conversion type string smi _ convert, wherein if a substring of smi _ sub exists in smi, the substring is partially converted into the smi _ convert; the smi _ sub substring comprises at least one site [ Ri ] connected with the rest part of the smi, the smi _ covert comprises the same site [ Ri ] at the corresponding position, the smi _ covert also comprises atoms to be cut off, and the cutting process is as follows:

(1) reading smi and smi _ sub, converting the smi _ sub into a graph structure mol and a graph structure mol _ sub, generating an atom mapping relation table maplist2 from the mol _ sub to the mol, and if the mol _ sub substructure does not exist in the mol, not performing segmentation; if a plurality of mol _ sub substructures exist in the mol, a plurality of mapping relations are stored in the maplist 2; whether atoms corresponding to the connection sites meet the limitation of the [ Ri ] type in the smi _ sub is considered during mapping, and if not, the mapping fails; when smi _ sub is read in, the position of the connection site [ Ri ] is saved;

(2) reading smi _ convert, converting the smi _ convert into a graph structure mol _ convert, generating an atom mapping relation table maplist1 from mol _ convert to mol _ sub, finding the position cut _ idx of the cut atom [ ], and finding the position cut _ connect _ idx of the atom connected with the cut atom [ ]; for each atom in mol _ convert, when the corresponding atom cannot be found in mol _ sub, representing the mapping atom position of the atom by-1;

(3) for each mapping relation in maplist2, a segmentation result is obtained:

(3.1) adding the mol and the mol _ sub to obtain updated mol;

(3.2) judging whether the mapping relation can cause the ring structure in the mol to be cut, if so, skipping the mapping relation, and executing the next mapping relation, otherwise, executing the step 3.3;

(3.3) collecting the positions of atoms which are mapped to the non-excision in the mol, namely the positions corresponding to the elements of > -1 in maplist1, and storing the positions to the not _ cut _ idxs through the mapping relation;

collecting and mapping the position of the atom connected with the cut atom [ ] in the mol, namely the atom corresponding to cut _ connect _ idx, and storing the atom to cut _ connect _ idxs;

collecting positions of all cut atoms mapped to mol, namely positions corresponding to elements which are-1 in maplist1, and storing the cut atoms to cut _ idxs, and storing atoms which are not in the not _ cut _ idxs in adjacent atoms to the cut points to cut _ idxs; maintaining a cut _ idxs list at each excision position, and if the corresponding atom in the mol cannot be found at the position, the position list is empty;

for cut _ idxs of each cutting position, if the cut _ idxs is empty, cutting a hydrogen atom on the position, and if the position does not have the hydrogen atom, skipping the cutting of the mapping relation and executing the next mapping relation; if not, collecting atoms to be cut off at the position, including the atom at the position and the atoms adjacent to the atom and not in cut _ connect _ idxs, and recursively collecting adjacent atoms of the adjacent atoms at the position;

collecting the positions of atoms to be cut in mol, namely those atoms in cut _ connect _ idx, including the atoms of the positions and the atoms adjacent to the positions but not in cut _ connect _ idx and cut _ idx, wherein the atoms have new positions in new mol, and recursively collecting the adjacent atoms of the positions;

(3.4) linking atom: connecting cut _ connect _ idxs with corresponding atoms in cut _ idx by chemical bonds;

(3.5) copy key attribute: copying bond attributes from the mol _ sub to newly added bonds in the mol; newly added atoms are collected, namely atoms at two ends of a newly added bond;

(3.6) updating the \ and/or spatial attributes of the double bond: if the mol spatial attribute structure contains atoms to be cut, the spatial attribute needs to be updated; atoms at two ends of the double bond and atoms connected with the atoms are stored in the double bond space attribute structure, if bonds exist between the atoms at two ends in the double bond space attribute structure and the newly added atoms in the previous step, the corresponding atoms in the structure are replaced by the newly added atoms;

(3.7) cutting the collected atoms to be cut off;

(3.8) outputting a sliced molecule of mapplist 2 mapping relation.

2. The method for chemical formula splitting according to claim 1, characterized in that, according to the characteristics of each reaction type, the atom type represented by the site [ Ri ] in the reaction type string smi _ sub is defined, i.e. the type of atom [ Ri ] corresponding to the formula to be split is defined, including the site [ Ri ] being not hydrogen (H), only C, N, O atom, or being any atom, defined according to the smart specification.

3. The method according to claim 1, wherein in step (1), the formula is read in a format of Canonical SMILES, and the other format of the input is converted into the format.

4. The method for splitting chemical formula according to claim 1, wherein in step (1), the mapping is performed by considering whether the atoms corresponding to the connecting sites satisfy the restriction on the type of [ Ri ] in smi _ sub, and if not, the mapping fails.

5. The method for splitting chemical molecular formula according to claim 1, wherein in the step (3.2), it is determined whether the mapping relationship would result in the ring structure in mol being split, that is, it is determined whether mol _ sub maps to a part of the ring structure in mol, specifically: the corresponding atom A of the to-be-cut site [ Ri ] in mol is found through the mapping relationship in maplist2, if A has an adjacent atom A _ neighbor not in the list to be cut off and A _ neighbor is on the ring, the ring will be cut.

6. The method for splitting chemical formulas according to claim 1, wherein in the step (3.5), the bond attributes are copied from mosub to newly added bonds in mol, that is, atoms at two ends of each bond in mosub are found, two atoms corresponding to mol are found, and the bond attribute between the two atoms is updated to the attribute value of the corresponding bond in mosub.

Images