A hierarchical screening method, device, system and medium for 3D compound similarity

Abstract

The present invention discloses a hierarchical screening method, device, system and medium for 3D compound similarity. The method comprises the following steps: inputting three-dimensional information of a query compound; preprocessing the query information of the query compound, comparing the query compound with compounds in a molecular library, screening the search range in the molecular library, and reducing the input scale; and calculating the similarity between the query compound and molecules in the molecular library based on the preprocessed search range and query information.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided drug research and development, and particularly to a hierarchical screening method, device, system and medium for 3D compound similarity. Background Art

[0002] Molecular similarity is a key concept in drug discovery. It is based on the assumption that molecules with similar structures usually have similar properties. Evaluating the similarity between small molecules is very effective in drug discovery and development. Among various small molecule 3D similarity calculation methods, the evaluation of shape similarity has received increasing attention not only in virtual screening but also in various applications such as molecular target prediction and drug repurposing. Currently, a variety of methods have been developed to describe molecular shape and determine the shape similarity between small molecules.

[0003] The methods include the method based on "atomic distance" and the representation of "Gaussian superposition based on atomic centers".

[0004] The method based on "atomic distance" has great advantages in terms of calculation speed, but it is relatively difficult to visually inspect the shape similarity, especially for molecules with relatively low structural similarity.

[0005] The "Gaussian-based shape similarity method centered on atoms" has many advantages over other shape similarity methods. However, it requires much more computing resources compared to the method based on "atomic distance". The main advantage of the Gaussian-based shape similarity method centered on atoms is visualization. Visualization of the shape similarity between two molecules is very helpful for deriving and optimizing the structure-activity relationship.

[0006] Therefore, how to quickly and accurately complete similarity queries in data of billions or tens of billions has become an urgent problem to be solved. Summary of the Invention

[0007] The object of the present invention is to provide a hierarchical screening method, device, system and medium for 3D compound similarity, which is used to improve the efficiency of query screening for 3D similarity of small molecule compounds at the tens of billions level.

[0008] The technical solutions adopted by the present invention to solve the above technical problems are as follows:

[0009] A hierarchical screening method for 3D compound similarity includes the following steps:

[0010] Input the three-dimensional information of the query compound;

[0011] Preprocess the query information of the query compound, compare the query compound with the compounds in the molecular library, and screen the search range in the molecular library to reduce the input scale;

[0012] Calculate the similarity between the query compound and the molecules in the molecular library based on the preprocessed search scope and query information.

[0013] Further, the input of the query compound includes the following steps:

[0014] Input the three-dimensional information of the query compound, where the three-dimensional information includes the type of each atom in the query compound and its coordinate values.

[0015] Perform validity verification on the input information, and issue a query request if the verification passes.

[0016] Further, the preprocessing of the input query information includes the following steps:

[0017] According to the compound volume similarity threshold inequality N*T <= Q <= N / T, where T is the similarity threshold, Q is the Gaussian volume of the query compound, and N is the Gaussian volume of the compound under the query condition;

[0018] Compare the query compound with the compounds in the molecular library, filter out the compounds that do not satisfy the inequality to narrow the search scope of the molecular library, and reduce the input scale for primary screening.

[0019] Further, the preprocessing of the input query information includes the following steps:

[0020] Coarsely classify the molecular shape according to the PMI value, and coarsely classify the query compound into simple geometric models such as rectangles, rhombuses, and circles;

[0021] Compare the query compound with the molecules in the molecular library, eliminate the molecules with a huge geometric difference, narrow the search scope of the molecular library, and reduce the input scale for secondary screening.

[0022] Further, the tertiary screening based on the comparison of the atomic distances of the query compound with the data in the molecular library includes the following steps:

[0023] Based on the three-dimensional information of the query compound, obtain the distances between the atoms in the query compound to form a query distance matrix.

[0024] Compare the similarity of the query distance matrix with the distance matrix of the molecules in the molecular library;

[0025] Eliminate the molecules with a similarity lower than the threshold, narrow the search scope of the molecular library, and reduce the input scale for tertiary screening.

[0026] Further, the calculation of the similarity between the query compound and the molecules in the molecular library based on the preprocessed search scope and query information includes the following steps:

[0027] Based on the preprocessed search scope and query information, a calculation request is issued;

[0028] Meanwhile, multiple computing platforms are launched for parallel computing to calculate the Gaussian superposition between the query compound and the molecules in the molecular library, so as to calculate the 3D similarity.

[0029] Furthermore, the computing platform is a GPU computing node.

[0030] Another embodiment of the present invention also provides a hierarchical screening system for 3D compound similarity, and the device includes:

[0031] A query input module for inputting the three-dimensional information of the query compound;

[0032] An information preprocessing module for preprocessing the query information of the query compound, comparing the query compound with the compounds in the molecular library, screening the search scope in the molecular library, and reducing the input scale;

[0033] A similarity calculation module for calculating the similarity between the query compound and the molecules in the molecular library based on the preprocessed search scope and query information.

[0034] Another embodiment of the present invention also provides a hierarchical screening device for 3D compound similarity, and the system includes at least one processor; and,

[0035] A memory communicatively connected to the at least one processor; wherein,

[0036] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned hierarchical screening method for 3D compound similarity.

[0037] Another embodiment of the present invention also provides a hierarchical screening storage medium for 3D compound similarity. The non-volatile computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by one or more processors, the one or more processors can execute the above-mentioned hierarchical screening method for 3D compound similarity.

[0038] Applying the technical solution of the present invention, at the order of magnitude of tens of billions, based on the threshold of the volume similarity of compounds, small molecule compounds that do not meet the conditions are clearly filtered out in the similarity query; at the same time, a pre-screening of small molecule compounds is carried out by using the PMI rough shape filtering method; combined with the advantage of fast calculation speed of the method based on atomic distance, the small molecule compounds are preprocessed. By preprocessing the query information of the query compound, the query compound is compared with the compounds in the molecular library, the search range in the molecular library is screened, the input scale is reduced, and thus the calculation amount is reduced and the calculation efficiency is improved; finally, the Gaussian overlap between the query compound and the molecules in the molecular library is calculated to compare the 3D similarity.

[0039] Other features and advantages of the invention will be described in the following specification, and in part will be obvious from the specification, or will be understood by practicing the present invention. The objectives and other advantages of the present invention can be realized and obtained by the structure specifically pointed out in the written specification and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The present invention will be described in detail below with reference to the drawings to make the above advantages of the present invention more clear.

[0041] Figure 1 is a flowchart of a hierarchical screening method for 3D compound similarity of the present invention;

[0042] Figure 2 is a schematic diagram of an embodiment of a virtual platform for a hierarchical screening method for 3D compound similarity of the present invention;

[0043] Figure 3 is a schematic diagram of functional modules of an embodiment of a hierarchical screening device for 3D compound similarity of the present invention

[0044] Figure 4 is a schematic diagram of the hardware structure of an embodiment of a hierarchical screening device for 3D compound similarity of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] To make the objectives, technical solutions and effects of the present invention clearer and more definite, the present invention will be further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0046] As Figure 1-2 shown, a hierarchical screening method for 3D compound similarity includes the following steps:

[0047] S100. Input the three-dimensional information of the query compound;

[0048] The three-dimensional structure information includes the type and coordinate values of each atom in the query compound; the van der Waals radius corresponding to each atom is obtained according to the type of atom in the query compound, and the three-dimensional structure information is converted into a set of Gaussian spheres representing the atoms in the query compound. The radius of each Gaussian sphere is the same as the van der Waals radius of the corresponding atom, and the position of each Gaussian sphere is the same as the coordinates of the corresponding atom; the volume of the query compound is represented by a Gaussian function

[0049] S200. Preprocess the query information of the query compound, compare the query compound with the compounds in the molecular library, screen the search range in the molecular library, and reduce the input scale

[0050] Preprocess the query information of the query compound through three-dimensional information such as the number of atoms, atomic distance, molecular shape, and overlapping volume between atoms, perform hierarchical screening of the search range, reduce the input scale, thereby reducing the calculation amount and improving the retrieval efficiency

[0051] S300. Calculate the similarity between the query compound and the molecules in the molecular library based on the preprocessed search range and query information

[0052] Compared with the initial query information input scale and retrieval range, the preprocessed search range and query information have been greatly reduced. Finally, only one-by-one comparison is performed within a small range, and the 3D Gaussian superposition between the query compound and the molecular compounds in the molecular library is calculated

[0053] While ensuring the visualization and accuracy of the 3D similarity query of molecules, the screening efficiency at the order of tens of billions is effectively improved. Especially for big data platforms, for multiple molecular library data, the calculation amount of screening is huge. The hierarchical screening method not only improves the efficiency but also has visualization and high accuracy

[0054] In this embodiment, the steps for inputting the query compound are as follows

[0055] Input the three-dimensional information of the query compound, where the three-dimensional information includes the type and coordinate values of each atom in the query compound

[0056] The van der Waals radius of an atom can be obtained through the type of atom, and then the volume of each atom can be obtained. The distance between atoms and the rough shape of the query compound can be calculated through the coordinates of the atoms

[0057] Perform validity verification on the input information. If the verification passes, send a query request

[0058] Before enabling the computing power query, first perform validity verification on the input retrieval information to ensure that the retrieval formula can be generated normally, avoid wasting computing power resources, especially for big data platforms

[0059] In this embodiment, the preprocessing of the input query information includes the following steps:

[0060] According to the compound volume similarity threshold inequality N*T <= Q <= N / T (where T is the similarity threshold, Q is the Gaussian volume of the query compound, and N is the Gaussian volume of the compound under the query condition;

[0061] Compare the query compound with the compounds in the molecular library, filter out the compounds that do not satisfy the inequality to narrow the search scope of the molecular library, and reduce the input scale for primary screening.

[0062] Based on the atomic species and coordinates of the query compound, the volume of the query compound can be calculated. Compare the volume of the query compound with the molecules in the molecular library to identify and filter out the small molecule compounds that do not meet the conditions in the similarity query, narrow the search scope of the molecular library, and reduce the input scale for primary screening. Thus, the purpose of reducing the calculation amount and improving the filtering efficiency is achieved. Moreover, the calculation of the volume of the query compound and the comparison calculation amount are relatively small, and the computing power requirement is low. Through primary screening, the molecules that are obviously not in line in the molecular library can be quickly filtered out, reducing the input scale.

[0063] In this embodiment, the preprocessing of the input query information includes the following steps:

[0064] Coarsely classify the molecular shape according to the PMI value, and coarsely classify the query compound into simple geometric models such as rectangles, rhombuses, and circles;

[0065] Compare the query compound with the molecules in the molecular library, and eliminate the molecules with a large geometric difference to narrow the search scope of the molecular library and reduce the input scale for secondary screening.

[0066] Based on the coordinate positions and atomic volume sizes of each atom in the query compound, the query compound can be coarsely classified into simple geometric models such as rectangles, rhombuses, and circles;

[0067] On the premise of the primary screening based on volume, perform secondary screening through the geometric shape of the query compound and the molecular shape in the molecular library, and filter out the molecules with a large difference in molecular shape. Further reduce the input scale, reduce the input scale and calculation amount for the final Gaussian superposition calculation, and improve the screening efficiency.

[0068] In this embodiment, the tertiary screening based on the comparison of the atomic distances of the query compound with the data in the molecular library includes the following steps:

[0069] Based on the three-dimensional information of the query compound, obtain the distances between the atoms in the query compound to form a query distance matrix,

[0070] The query distance matrix is compared with the distance matrices of the molecules in the molecular library for similarity.

[0071] Molecules with similarity lower than the threshold are removed, narrowing the search scope of the molecular library and reducing the input scale for the third-level screening.

[0072] In the third-level screening, only the spatial distances between atomic components in the molecule are considered, without considering the types of atoms. The third-level screening algorithm is simple and easy to implement. In the query compound, the distances between pairwise atoms are obtained and recorded one by one to form the query distance matrix A, and the query distance matrix is compared with the distance matrix B of the molecules in the molecular library. Compound molecules that theoretically cannot satisfy the inequality are filtered to reduce the input scale.

[0073] In this embodiment, calculating the similarity between the query compound and the molecules in the molecular library based on the preprocessed search scope and query information includes the following steps:

[0074] Based on the preprocessed search scope and query information, a calculation request is issued.

[0075] Multiple computing platforms are simultaneously launched for parallel computing to calculate the Gaussian superposition between the query compound and the molecules in the molecular library, thereby calculating the 3D similarity algorithm.

[0076] In the application of the virtual screening platform, after multiple preprocessings, the input scale of the backend 3D similarity calculation is greatly reduced. Utilizing the advantages of parallel computing on the distributed platform, multiple computing platforms are simultaneously launched, and the Gaussian-based shape similarity method centered on atoms is used for the final 3D similarity calculation. Parallel computing by multiple computing platforms greatly improves the calculation speed and screening efficiency.

[0077] In this embodiment, the computing platform is a GPU computing node. The cores of the GPU are good at completing tasks with simple control logic, focusing on computing and parallelism. Compared with the CPU, the GPU has a large amount of computing power. When the query compound is compared with the molecular compounds in the molecular library, the advantage of parallel computing of the GPU is utilized, greatly improving the efficiency of calculating the 3D Gaussian superposition between the query compound and the compounds in the library and shortening the retrieval time.

[0078] Based on the above technical solutions, to solve the 3D similarity query of large-scale small molecule compounds, a calculation implementation method based on GPU is proposed to perform similarity calculation based on Gaussian superposition centered on atoms, improving the calculation speed by an order of magnitude; at the same time, based on the big data parallel computing architecture, distributed multi-GPU node parallel computing is realized, and the calculation speed is increased by another order of magnitude; greatly accelerating the molecular similarity query step in virtual screening.

[0079] At the order of magnitude of tens of billions, we propose an inequality based on the similarity threshold of the volume of chemical compounds: N*T <= Q <= N / T (where T is the similarity threshold, Q is the Gaussian volume of the query compound, and N is the Gaussian volume of the compound under the query condition). This inequality is used to clearly filter out small molecule compounds that do not meet the conditions in similarity queries. At the same time, a rough shape filtering method based on PMI is adopted for the pre-screening of small molecule compounds, and then combined with the advantage of fast calculation speed of the method based on "atomic distance" to preprocess small molecule compounds. A hierarchical and distributed 3D similarity calculation solution based on "Gaussian shape" is proposed.

[0080] Another embodiment of the present invention also provides a hierarchical screening system for 3D compound similarity. The device includes:

[0081] A query input module 10 for inputting the three-dimensional information of the query compound;

[0082] An information preprocessing module 20 for preprocessing the query information of the query compound, comparing the compound to be queried with the compounds in the molecular library, screening the search range in the molecular library, and reducing the input scale;

[0083] A similarity calculation module 30 for calculating the similarity between the query compound and the molecules in the molecular library based on the preprocessed search range and query information.

[0084] Another embodiment of the present invention provides a system for generating a three-dimensional model, as Figure 3 shown. The system 50 includes:

[0085] One or more processors 510 and a memory 520. Figure 3 Taking one processor 510 as an example for introduction, the processor 510 and the memory 520 can be connected through a bus or other means. Figure 3 Taking the connection through the bus as an example.

[0086] The processor 510 is used to complete various control logics of the system 50. It can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine), or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination of these components. Additionally, the processor 510 can also be any traditional processor, microprocessor, or state machine. The processor 510 can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP and / or any other such configuration.

[0087] The memory 520, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as program instructions corresponding to the hierarchical screening method for 3D compound similarity in the embodiments of the present invention. The processor 510 executes various functional applications and data processing of the system 50 by running the non-volatile software programs, instructions, and units stored in the memory 520, that is, implements the hierarchical screening method for 3D compound similarity in the above method embodiments.

[0088] The memory 520 may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the system 50, etc. In addition, the memory 520 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the memory 520 may optionally include a memory remotely provided with respect to the processor 510, and these remote memories can be connected to the system 50 through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0089] One or more units are stored in the memory 520 and, when executed by one or more processors 510, execute the hierarchical screening method for 3D compound similarity in any of the above method embodiments. For example, execute the method steps S100 to step S400 described above. Figure 1 in the above.

[0090] As Figure 4 shown, the embodiments of the present invention provide a non-volatile computer-readable storage medium. The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are executed by one or more processors. For example, execute the method steps S100 to step S300 described above. Figure 1 in the above.

[0091] By way of example, the non-volatile storage medium can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM), or flash memory. The volatile memory can include random access memory (RAM) as an external cache memory. By way of illustration and not limitation, RAM can be obtained in many forms such as synchronous RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The disclosed memory components or memories of the operating environment described herein are intended to include one or more of these and / or any other suitable types of memories.

[0092] Another embodiment of the present invention provides a computer program product. The computer program product includes a computer program stored on a non-volatile computer-readable storage medium. The computer program includes program instructions that, when executed by a processor, cause the processor to execute the hierarchical screening method for 3D compound similarity in the above method embodiments. For example, execute the method steps S100 to step S400 described above. Figure 1 in the method.

[0093] In summary, the present invention discloses a hierarchical screening method, device, system, and medium for 3D compound similarity. While lightweighting the bone model, the method also binds the bone model to the skin model and provides tools such as bone fine-tuning, action fine-tuning, and skin fine-tuning to create natural and smooth three-dimensional virtual objects, strictly controlling the volume and transmission amount of data, greatly reducing the loading waiting time and the program's budget amount, and finally saving the file as a general-purpose file, which is applicable to different three-dimensional model application platforms and has strong versatility.

[0094] The embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0095] Through the descriptions of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform. Of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the related technology can be embodied in the form of a software product. This computer software product can exist in a computer-readable storage medium, such as ROM / RAM, magnetic discs, optical discs, etc., and includes several instructions for causing a computer electronic device (which can be a personal computer, a server, or a network electronic device, etc.) to execute the methods of each embodiment or some parts of the embodiments.

[0096] Among other things, conditional language such as "can", "be able to", "may", or "could", unless specifically stated otherwise or otherwise understood within the context in which it is used, generally is intended to convey that a particular embodiment can include (while other embodiments do not include) a particular feature, element, and / or operation. Thus, such conditional language generally is also intended to imply that the feature, element, and / or operation is in some way required for one or more embodiments or that one or more embodiments must include logic for determining whether the feature, element, and / or operation is included or will be performed in any particular embodiment in the presence or absence of input or prompting.

[0097] What has been described herein in the specification and the drawings includes examples of a method, apparatus, system, and medium for providing a hierarchical screening of 3D compound similarity. Of course, it is not possible to describe every conceivable combination of elements and / or methods for the purpose of describing the various features of the present disclosure. However, it can be recognized that many additional combinations and permutations of the disclosed features are possible. Therefore, it is apparent that various modifications can be made to the present disclosure without departing from the scope or spirit of the present disclosure. Additionally, or in the alternative, other embodiments of the present disclosure may be apparent from consideration of the specification and the drawings and from practice of the present disclosure as presented herein. The intention is that the examples presented in the specification and the drawings be considered illustrative in all respects and not restrictive. Although specific terms have been employed herein, they are used in a generic and descriptive sense and not for purposes of limitation.

[0098] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A hierarchical screening method for 3D compound similarity, characterized in that, it includes the following steps: Input the three-dimensional information of the query compound; Preprocess the query information of the query compound. According to the compound volume similarity threshold inequality N*T <= Q <= N / T, where T is the similarity threshold, Q is the Gaussian volume of the query compound, and N is the Gaussian volume of the compound under the query condition; Compare the query compound with the compounds in the molecular library, filter out the compounds that do not satisfy the inequality to narrow the search range of the molecular library, and reduce the input scale for primary screening; Coarsely classify the molecular shape according to the PMI value, coarsely classify the query compound into simple geometric models such as rectangular, rhombic, and circular, compare the query compound with the molecules in the molecular library, eliminate the molecules with huge geometric differences, narrow the search range of the molecular library, and reduce the input scale for secondary screening; Based on the three-dimensional information of the query compound, obtain the distances between atoms in the query compound to form a query distance matrix, and compare the similarity of the query distance matrix with the distance matrix of the molecules in the molecular library; Eliminate the molecules with similarity lower than the threshold, narrow the search range of the molecular library, and reduce the input scale for tertiary screening; Based on the preprocessed search range and query information, calculate the similarity between the query compound and the molecules in the molecular library.

2. The hierarchical screening method for 3D compound similarity according to claim 1, characterized in that, the input of the query compound includes the following steps: Input the three-dimensional information of the query compound, and the three-dimensional information includes the type and coordinate values of each atom in the query compound; Verify the validity of the input information, and send a query request if the verification passes.

3. The hierarchical screening method for 3D compound similarity according to claim 1, characterized in that, the calculation of the similarity between the query compound and the molecules in the molecular library based on the preprocessed search range and query information includes the following steps: Based on the preprocessed search range and query information, send a calculation request; Simultaneously start multiple computing platforms for parallel computing, calculate the Gaussian superposition between the query compound and the molecules in the molecular library, so as to calculate the 3D similarity calculation.

4. The hierarchical screening method for 3D compound similarity according to claim 3, characterized in that, the computing platform is a GPU computing node.

5. A hierarchical screening device for 3D compound similarity, used to run the hierarchical screening method for 3D compound similarity according to any one of claims 1-4, characterized in that, it includes: A query input module, used to input the three-dimensional information of the query compound; An information preprocessing module, used to preprocess the query information of the query compound, compare the query compound with the compounds in the molecular library, screen the search range in the molecular library, and reduce the input scale; A similarity calculation module; Used to calculate the similarity between the query compound and the molecules in the molecular library based on the preprocessed search range and query information.

6. A hierarchical screening system for 3D compound similarity, characterized in that, it includes at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the hierarchical screening method for 3D compound similarity according to any one of claims 1-4.

7. A hierarchical screening storage medium for 3D compound similarity, characterized in that the storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by one or more processors, the one or more processors can be enabled to execute the hierarchical screening method for 3D compound similarity according to any one of claims 1-4.

Images