Multi-instance GPU-based system for discovering drug candidate
A drug candidate discovery system may include: a project management module configured to create a project for adding a task to perform drug candidate discovery; a simulation management module configured to create a simulation on the created project; a simulation setting module configured to set a simulation workflow for the simulation based on input from a user, using a canvas area and a simulation setting area containing a protein structure data input area and a task module selection area, each containing one or more objects that can be dragged and dropped onto the canvas area and converted into a node; a simulation workflow management module configured to manage information on nodes that can precede or follow in the simulation workflow; and a simulation execution module configured to manage and execute a task for running the simulation workflow by dividing and allocating GPU resource to the respective nodes.
1 . A drug candidate discovery system, comprising:
at least one computing device including at least one processor and at least one memory storing instructions that, when executed by the processor, cause the computing device to perform operations comprising:
creating a project for adding a task to perform drug candidate discovery;
creating a simulation desired by a user on the created project;
setting a simulation workflow for the simulation based on input from a user, using a canvas area and a simulation setting area comprising a protein structure data input area and a task module selection area, each comprising one or more objects that can be dragged and dropped onto the canvas area and converted into a node;
managing information on nodes that can precede or follow in the simulation workflow;
managing and executing a task for running the simulation workflow by dividing and allocating GPU resource of the computing device to the respective nodes; and
providing a plurality of tasks for performing drug candidate discovery, corresponding to elements of the simulation workflow set by dragging and dropping onto the canvas area by a user,
wherein:
the GPU resource comprises a virtual GPU,
the virtual GPU is assigned a first capacity representing the capacity supported by the virtual GPU,
for each of the plurality of tasks, a second capacity representing the capacity required during task execution is assigned,
for each of the plurality of tasks, the first capacity is compared with the second capacity to determine executability, and
the second capacity is modified to a different value by the user.
2 . The drug candidate discovery system of claim 1 , wherein the operation of managing and executing the task comprises:
scheduling an execution of the task using a task queue; and
managing GPU based on a multi-instance GPU using a Docker container and executing the scheduled task by utilizing a virtual GPU generated based on the GPU.
3 . The drug candidate discovery system of claim 2 , wherein the operation of providing the plurality of tasks comprises:
providing the task stored in the form of the Docker container to the operation of managing and executing the task.
4 . The drug candidate discovery system of claim 3 , wherein the operation of providing the plurality of tasks comprises:
allocating to the Docker container and providing to the operation of managing and executing the task: a first task for automatically identifying an optimal docking site in a target protein structure, a second task for predicting the tertiary structure of a protein from an amino acid sequence, a third task for analyzing and sorting an actual binding energy (kcal/mol) and providing the analyzed binding energy to a user, a fourth task for converting kcal/mol into Kd/Ki (μM) for selected ligands and performing comparative analysis, and a fifth task for predicting absorption, distribution, metabolism, excretion, and toxicity based on the chemical structure of ligands.
5 . The drug candidate discovery system of claim 1 , wherein the operations further comprise:
displaying on a screen:
a first user interface for receiving, from a user, the number of threads to be used to execute the task on the virtual GPU, and
a second user interface for displaying, on the screen, the number of required tokens that a user must pay to execute the task, wherein the number of tokens increases or decreases based on the number of threads.
6 . The drug candidate discovery system of claim 1 , wherein the operations further comprise:
displaying a screen providing a list of the virtual GPU to the user, and
wherein the list displayed on the screen includes: a name identifying the virtual GPU, information on a task module available for the virtual GPU, and a maximum capacity supported by the virtual GPU.
7 . The drug candidate discovery system of claim 1 , wherein the protein structure data input area comprises one or more objects related to a function of uploading protein structure data, and
wherein the one or more objects can be dragged and dropped onto the canvas area and converted into: a first node configured to receive protein structure data in the form of a PDB (Protein Data Bank) file from a user, a second node configured to receive protein structure data in the form of a PDB code from a user, a third node configured to receive protein structure data in the form of a protein sequence file from a user, or a fourth node configured to receive protein structure data in the form of a protein sequence from a user.
8 . The drug candidate discovery system of claim 7 , wherein the task module selection area comprises one or more objects related to functions for performing detailed tasks in drug candidate discovery based on the uploaded protein structure data, and
wherein the one or more objects can be dragged and dropped onto the canvas area and converted into: a fifth node configured to identify an optimal docking site in a target protein structure, a sixth node configured to predict the tertiary structure of a protein from an amino acid sequence, a seventh node configured to analyze and sort actual binding energy (kcal/mol) and provide the analyzed binding energy to a user, or an eighth node configured to convert kcal/mol into Kd/Ki (μM) for selected ligands and perform comparative analysis.
9 . The drug candidate discovery system of claim 8 , wherein nodes that can precede the fifth node comprise the first node, the second node, the third node, the fourth node, and the sixth node, and
wherein nodes that can follow the fifth node comprise the seventh node.
10 . The drug candidate discovery system of claim 8 , wherein nodes that can precede the sixth node comprise the third node and the fourth node, and
wherein nodes that can follow the sixth node comprise the fifth node.
11 . The drug candidate discovery system of claim 8 , wherein nodes that can precede the seventh node comprise the fifth node, and
wherein nodes that can follow the seventh node comprise the eighth node.
12 . The drug candidate discovery system of claim 1 , wherein a node connection shape is displayed on the node placed in the canvas area, and
while a user clicks on a node connection shape displayed on the right side of a certain node, the color or shape of a node connection shape displayed on the left side of other connectable node changes.
13 . The drug candidate discovery system of claim 12 , wherein the operation of managing information on nodes comprises:
managing information for determining whether connections between nodes are possible through metadata.
14 . A drug candidate discovery system, comprising:
at least one computing device including at least one processor and at least one memory storing instructions that, when executed by the processor, cause the computing device to perform operations comprising:
creating a project for adding a task to perform drug candidate discovery;
creating a simulation desired by a user on the created project;
setting a simulation workflow for the simulation based on input from a user, using a canvas area and a simulation setting area comprising a protein structure data input area and a task module selection area, each comprising one or more objects that can be dragged and dropped onto the canvas area and converted into a node;
managing information on nodes that can precede or follow in the simulation workflow;
managing and executing a task for running the simulation workflow by dividing and allocating GPU resource of the computing device to the respective nodes; and
displaying a first screen providing a list of a server to the user,
wherein the list displayed on the first screen includes: a name identifying the server, information on a virtual GPU used by the server, port information used to access the server, a total capacity supported by the server, a remaining capacity currently available on the server, and a usability status of the server.
15 . The drug candidate discovery system of claim 14 , wherein the operations further comprise:
displaying a second screen providing a list of a virtual GPU used by the server to a user, and
wherein the list on the second screen includes: a name identifying the server, GPU ID information of the virtual GPU used by the server, type information of the virtual GPU used by the server, a maximum capacity supported by the virtual GPU, a currently available capacity on the virtual GPU, and a usability status of the virtual GPU.
16 . A drug candidate discovery system, comprising:
at least one computing device including at least one processor and at least one memory storing instructions that, when executed by the processor, cause the computing device to perform operations comprising:
creating a project for adding a task to perform drug candidate discovery;
creating a simulation desired by a user on the created project;
setting a simulation workflow for the simulation based on input from a user, using a canvas area and a simulation setting area comprising a protein structure data input area and a task module selection area, each comprising one or more objects that can be dragged and dropped onto the canvas area and converted into a node;
managing information on nodes that can precede or follow in the simulation workflow; and
managing and executing a task for running the simulation workflow by dividing and allocating GPU resource of the computing device to the respective nodes; and
displaying a first screen providing a list of a task queue to a user,
wherein the list on the first screen includes: position information of a task inserted into the task queue, URL information associated with the task, information on whether the task inserted into the task queue are released, a name of the task inserted into the task queue, an ID of the task, and the number of threads used to execute the task.
17 . The drug candidate discovery system of claim 16 , wherein the operations further comprise:
determining whether the task inserted into the task queue is executable, and
executing the task if the task is determined to be executable,
wherein executability of the task is determined by comparing a capacity required for executing the task with a currently available capacity on the virtual GPU, and the task is determined to be executable if the capacity required for executing the task does not exceed the currently available capacity on the virtual GPU, and not executable if the capacity required for executing the task exceeds the currently available capacity on the virtual GPU.
18 . The drug candidate discovery system of claim 16 , wherein the operations further comprise:
displaying a second screen providing a list to a user, the list including information on a blocked capacity during the execution of the task that was scheduled in the task queue.