JP2024518336A - Protein-protein interaction detection - Google Patents

Abstract

A composite image is accessed, the composite image including a plurality of complex images of a sample of a protein-protein complex including a first protein and a second protein. The composite image is masked to generate a masked portion and an unmasked portion. A first three-dimensional (3d) shape of the first protein and a second 3d shape of the second protein are accessed. A plurality of docking models, each of which defines a candidate pose pair, are accessed. For each docking model, the first 3d shape, the second 3d shape and the candidate pose pair are applied, and a corresponding fit score is generated for the docking model that describes a goodness of fit between the pose pair and the docking model. Based on the fit score, one of the docking models is selected as a detected model for the protein-protein complex. [Selected Figure]

Description

This paper describes techniques for characterizing protein-protein binding using sensor data.

Protein-protein interactions (PPIs) are highly specific physical contacts established between two or more protein molecules as a result of interactions involving electrostatic forces, hydrogen bonds and hydrophobic effects. Many are physical contacts with chain-to-chain molecular relationships that occur within cells or living tissues in specific biomolecular contexts. Proteins rarely act alone, as their functions tend to be regulated. Many molecular processes within cells are carried out by molecular machines built from numerous protein components orchestrated by PPIs.

In immunology, an antigen (Ag) is a molecule or molecular structure that can be present on the outside of a pathogen that can be restricted by an antigen-specific antibody or a B cell antigen receptor. The presence of an antigen in the body usually triggers an immune response.

Techniques for characterizing protein-protein interactions are described herein. For example, the techniques can be useful in understanding how an antibody interacts with an antigen when developing molecules for clinical or biological use (e.g., drug development). Cryo-EM imaging of protein complexes can be performed and data from the imaging can be processed using computer systems to select docking models that describe the relative location, orientation, and binding of the two proteins.

One or more computer systems can be configured to perform a particular operation or action by having software, firmware, hardware, or a combination thereof installed on the system that, when operated, causes the system to perform the action. One or more computer programs can be configured to perform a particular operation or action by including instructions that, when executed by a data processing device, cause the device to perform the action. One general aspect includes a method for detecting protein-protein complex interactions: the method can include accessing a composite image including a plurality of complex images of a sample of a protein-protein complex including a first protein and a second protein. The method also includes accessing a first three-dimensional (3d) shape of the first protein and a second 3d shape of the second protein. The method also includes accessing a plurality of docking models, each of which defines a candidate pose pair. The method also includes applying, for each docking model, the first 3d shape, the second 3d shape, and the candidate pose pair to generate, for the docking model, a corresponding fit score that describes a goodness of fit between the pose pair and the docking model. The method also includes selecting one of the docking models as a detected model for the protein-protein complex based on the fit score. Other embodiments of this aspect include corresponding computer systems, apparatus, and computer programs recorded on one or more computer storage devices, each configured to perform the actions of the method.

The embodiment may include one or more of the following configurations. The method may include generating a plurality of complex images. Generating the composite image may include extracting a sub-image of the protein-protein-complex from the complex image; and orienting and classifying the sub-image. The complex image is a cryo-electron microscopy (cryo-EM) image. Each complex image may include a plurality of pixels, each having an address and holding a color value that represents a corresponding portion of the sample of the protein-protein complex. The composite image may include a plurality of pixels, each having an address and holding a color value that is a collection of color values of pixels having the same address in each of the plurality of complex images. Masking of the composite image is performed without specific user input. Masking of the composite image may include receiving a first user input that specifies a non-masking portion. Receiving a first user input that specifies a non-masking portion may include: generating a bounding box by connecting locations specified by the first user input; and recording a portion of the composite image within the bounding box as a non-masking portion. The first 3d shape is indexed as a first protein and the second 3d shape is indexed as a second protein. The first 3d shape is indexed as a first homolog of the first protein. The second 3d shape is indexed as a second homolog of the second protein. The candidate pose pair includes a candidate location, a candidate orientation, and a candidate docking area. The fitness score is a cross-correlation score between the composite image and the image generated by projecting the docking model onto the 2D space. Selecting one of the docking models as a detected model for the protein-protein complex based on the fitness score can include: identifying a subset of the docking models based on their corresponding fitness scores by one of the group consisting of: i) selecting docking models with top n fitness scores; and ii) selecting all docking models with fitness scores above a threshold m. Selecting one of the docking models as a detected model for the protein-protein complex can include receiving a second user input selecting one of the subset of docking models as the detected model. Implementations of the described techniques can include hardware, methods or processes, or computer software on a computer-accessible medium.

Implementations can include any, all, or none of the following advantages. This technique can advantageously use cryo-EM imaging results that are easier to obtain. For example, other techniques may require cryo-EM images of protein complex particles from a variety of angles. Unlike these techniques, this document describes a technique that can work with particle images from fewer angles. This is particularly advantageous for obtaining specific spatial orientations during sample preparation for cryo-EM due to the nature of some protein complexes. This is sometimes referred to as the challenge of preferred orientation. Other techniques may require extensive experimentation to discover how to overcome this challenge for a particular protein complex and may still fail. On the other hand, this technique can avoid the challenge all together. This can lead to a process that can be completed in a matter of hours or days, where some other processes take a matter of weeks or months to complete. In extreme cases, this technique may be the only recourse because, unlike other techniques, it works every time a protein particle image is obtained during cryo-EM imaging. Additionally, the technique can be advantageously configured to incorporate human user domain expertise in a process that is very rapid and requires little time. However, the technique can also be advantageously configured to proceed without the need for any specific human input at various stages, allowing for less human time and attention to complete the process.

Other configurations, aspects and potential advantages will become apparent from the accompanying description and drawings.

1 shows an exemplary system capable of detecting protein-protein interactions within a protein complex. Exemplary data are presented that can be used in detecting protein-protein complexes. 1 shows an exemplary process for detecting protein-protein complexes. 1 illustrates an exemplary process for creating a composite image. 1 illustrates an example process for masking a composite image. 1 illustrates an example process for selecting a detected model from a group of candidate models. 1A-1C show schematic diagrams of example computing devices and mobile computing devices.

Like reference numbers in the various drawings indicate like elements.

Protein-protein interactions can be visualized, for example, from cryo-EM images. A group of cryo-EM image fragments are aggregated into a single composite image, which is then masked to isolate the areas where the two proteins bind. The composite image is then sent to a group of docking models, each of which is scored to identify how well the model describes the docking shown in the composite image. Masking, docking and scoring can be applied several times (at least as many times as the number of proteins forming the PPI) to different parts of the protein complex. Based on this scoring, the best model is eventually identified.

For example, protein-protein interactions can be modeled using cryo-EM images or other types of images without the need to use single particle 3D reconstructions. Particles of protein complexes are extracted from a group of cryo-EM images, followed by alignment, classification and averaging. The averaged image of the protein complex is then masked to isolate individual proteins or some areas of the protein complex. The masked image is then searched against a series of 2D images projected consecutively from 3-D structures or models of individual protein components such as antigens and antibody fragment antigen binding (Fab). Cross-correlation is performed between the masked (or unmasked) image and the 2D projected images to identify the orientation of the antigen and/or Fab in the averaged 2D cryo-EM image. Protein-protein docking (antigen-Fab docking in the example) is performed using the cross-correlated identified interfaces as constraints for the binding site. In the case of Fab-antigen docking, we generate dozens of similar models (structural ensembles) of Fabs with diverse conformations of the CDRs. Each of the docking output poses was transformed into a series of 2D images by projection. Next, in the original 2D cryoEM, the averaged images and/or areas of the complex are masked out and then used as search templates to perform cross-correlation against the 2D images from the docking results. The docking result that gives the highest cross-correlation score for the masked complex is marked as the best model of the protein-protein complex. For complexes with two components, such as Fab-antigen, these steps complete the task. If there are further components to be identified, iterative masking and cross-correlation is performed.

FIG. 1 shows an exemplary system 100 that can detect protein-protein complex interactions. In the system 100, an imager 102 images a sample of a protein-protein complex 104. This can enable detection of the protein-protein complex 104 and enable the imager 102 to be a sensor that detects the docking characteristics of the protein complex 104.

Imager 102 is an imager capable of sensing physical phenomena of complex 104 and generating data (e.g., digital information) reflecting these physical phenomena. For example, the imager can be a cryo-electron microscope capable of measuring complex 104 by passing a beam of accelerated electrons through complex 104 to a sensor. Perturbations in the beam are recorded and measured to capture information about shape, cross-sectional density, etc. To aid in this sensing, complex 104 can be held in an ultra-thin layer of ice at low temperatures (e.g., cryogenic temperatures). As will be appreciated, complex 104 can include additional proteins and other components.

The imager 102 can generate the composite images 106. For example, the imager 102 can generate one set of composite images 106 for each composite 104. In some cases, some of the composite images 106 can be omitted, for example, to capture the composite 104 at an angle different from that of the other composites 104. As will be appreciated, some cryo-EM processes involve the composite being biased to a particular orientation during sensing of the grid, resulting in many, but possibly not all, composites 104 having the same or similar orientation. In addition, some of the composites 104 may not be captured in the corresponding image 106. For example, the imaging process may fail to capture the entire composite 104 (e.g., the composite biases to a particular orientation, which does not capture complete imaging of the composite from various orientations).

The composite images 106 can be aggregated into a composite image 108. The composite image 108 can represent the collection of composite images 106 in a format that allows for a single data object that is affected by each composite image and thus each detected composite 104.

The process of aggregating the composite images 106 into a composite image can include extracting, reorienting, and classifying elements of the composite images 106 so that the captured images of the composites 104 can be combined. For example, each of the composite images 106 can be examined to identify portions (e.g., groups of pixels) that exhibit background values and portions that exhibit the composites 104. The portions that exhibit the composites 104 can be extracted into a new data file and analyzed by computer vision to identify features such as unique clusters of values, the longest axis of the composites 104, etc. These extracted data can then be rotated to align the features, and thus the entire extracted data. For example, the data can be rotated by an angle such that the clusters of features have minimal error or difference relative to the template image, or the longest axis coincides with a particular angle (e.g., 0 or 90 degrees). The extracted data for each composite 104 can then be combined with the expectation that this will combine images of the composites 104 that are all in the same orientation and location in the working file.

The group of docking models 110 may each describe one possible model of protein-protein interaction. To characterize the protein-protein interactions of the complex 104, each docking model may be provided with i) a composite image 108, ii) an image mask 112 that masks out areas of the composite image 108 that are not expected to contain binding sites, iii) a 3D shape 114 that describes one of the proteins in the complex 104 in the data, and iv) a 3D shape 116 that describes another of the proteins in the complex 104 in the data. If more than two proteins are involved in PPI, additional masks and 3D shapes may be used. As will be appreciated, there may be cases where zero masks are used and where operations may be performed iteratively using a sequence of masks.

For each docking model, a score 118 is generated that describes a measure of fit between the docking model and the data provided thereto. In other words, the score 118 records how similar or different the model 110 is from the data 108, 112-116.

The maximum scores 118 can be examined to identify a selected docking model 120. For example, the two highest scores can be identified and rendered in a user interface of the computer 122. One of these docking models can then be selected, for example, by user input and/or a computer vision process.

In this example, the interaction of two proteins is described. However, it is understood that interactions with more than two proteins are possible. For example, the processes described herein can be repeated for each pair of proteins, or for each pair of proteins that contact each other. These repeated processes can be performed, for example, in series or in parallel.

FIG. 2 shows example data 106-114 that can be used in detecting protein-protein complexes. For example, the data 106-114 can include binary digital information stored in computer memory, transmitted over a data network between computing devices, etc. The data can be stored on disk in binary format and rendered on a display screen with colors and shapes defined by the binary data.

The complex images 106 may include cryo-electron microscopy (cryo-EM) images. Each image may include a bitmap of pixels, i.e., cells arranged in a regular two-dimensional grid, addressed by [x][y] to uniquely identify each cell. Each cell may include one or more values representing values in, for example, an intensity format that may use values between 0 and 1, a red-green-blue (RGB) format, a six-digit hex format, etc. The value of each pixel represents a corresponding portion of a sample of a protein-protein complex. For example, a sensor map of the image 102 may receive electrons passing through a portion of the protein-protein complex 104, convert the detection to a numerical value, and store the numerical value in a similarly addressed pixel in the composite image 106. As will be appreciated, the complex images 106 may be so named due to the recording of information about the complex 104.

The composite image 106 may include an aggregate (e.g., class average) of a group of composite images 106, e.g., various different composite images 106 of different instances of the same type of protein-protein complex 104 detected by the imager 102. Each image may include a bitmap of pixels, i.e., cells arranged in a regular two-dimensional grid, addressed by [x][y] to uniquely identify each cell. Each cell may include one or more values representing a color, e.g., in red-green-blue (RGB) format, 6-digit hex format, etc. The color value of each pixel represents a color value that is an aggregate of the color values of pixels having the same address in each of the multiple composite images. For example, for pixel [133][217] of the composite image 108, the color values of each pixel [133][217] in the group of extracted sub-images of the composite image 106 may be aggregated. This aggregation may be a simple average, summation, or other aggregation criterion appropriate to the data format of the pixel values and other technical coefficients.

The image mask 112 may include information specifying masked and unmasked portions of another image, such as the composite image 108. Each image may include a bitmap of pixels, i.e., cells arranged in a regular two-dimensional grid, addressed by [x][y] to uniquely identify each cell. Each cell may include one or more values representing a color, e.g., in red-green-blue (RGB) format, 6-digit hex format, etc. The color value of each pixel represents color values reserved for masked status, unmasked status, etc. For example, black and white colors may be used. Image 112' shows the image mask 112 superimposed on the composite image 108, where the masked section is rendered in black and the unmasked section is rendered using pixel values of the unmasked pixels of the composite image 108. In some configurations, the image mask 112 may include or use a bounding box that describes the edges of the masked or unmasked sections. For example, a process (e.g., user input selection, automated script) can identify groups of vertices and create edges between the vertices to create polygons that act as bounding boxes.

The 3D shape 114 may include information for specifying the shape of a single protein or other molecular structure. For example, the 3D shape 114 may include a Protein Data Bank (.pdb) file that records HEADER, TITLE, and AUTHOR records; REMARK records; SEQRES records; ATOM records; and HETATM records. However, other file types and other data models may be used. For example, the 3D shape 114 may include a Macromolecule Crystal Information File (.mmCIF) file that records data in a tag-value format for representing macromolecule structural data.

The 3D shapes 114 and 116 can be selected for use based on matching of the two proteins in the protein-protein complex 104. For example, if a first protein is known and has a fully described 3D shape 114, and a second protein is also known and has a fully described 3D shape 116, these 3D shapes 114 and 116 can be indexed by the name of the protein and used in these processes. However, in some cases, the 3D shapes of homologs of one or both of the proteins may be used. In such cases, structurally similar proteins can be identified as homologs, and the 3D shapes indexed by the name of the homolog can be accessed.

The docking model 110 includes structured data that defines possible pose pairs of two proteins in a protein-protein complex. For example, the pose pairs can include relative locations, relative orientations, and docking areas. The data can be organized to assume that a point in one protein is at point [0][0][0] in 3D space. The pose can then specify a translation (e.g., translation) in terms of [x][y][z] that defines the translation from the origin required to locate the second protein. The pose can also specify a rotation (e.g., spin) in terms of [x][y][z] that defines the rotation from the orientation of the first protein required to locate the orientation of the second protein. The docking area can specify one or more surfaces of the proteins that the model designates as docking surfaces where the two proteins dock or contact. The docking model 110 can be computationally generated according to expected rules that are believed to represent physical protein contact areas. The docking model 110 can be experimentally generated according to experiments that measure real-world samples of actual protein-protein complexes.

FIG. 3 illustrates an example process 300 for detecting protein-protein complexes. For example, process 300 can be performed with elements of system 100, and for clarity, the examples herein are described in terms of elements of system 100. However, other systems may be used to perform process 300 and other similar processes.

A composite image is accessed (302), the composite image including multiple complex images of a sample of a protein-protein complex including a first protein and a second protein. For example, the computer 122 may access the composite image 108 from an internal memory or from a remote (e.g., cloud-hosted) memory service. This may occur, for example, as a result of receiving a user input requesting analysis of a protein complex 104 imaged by the imager 102.

The composite image is masked (304) to generate masked and unmasked portions. For example, an image mask 112 may be applied to the composite image to designate masked and unmasked portions based on pixel values stored in the image mask 112. In some configurations, the image mask 112 is generated by an automated script or otherwise without specific user input. In some cases, computer vision techniques may be applied to identify features in the composite image 108, and the mask is generated by an automated computer vision process. In some configurations, the image mask 112 is generated using input from a user. In some cases, no masking of the composite image is performed. An example of one such process is described later in this document.

A first three-dimensional (3D) shape of the first protein and a second 3D shape of the second protein are accessed (306). For example, the computer 122 can access the 3D shapes 114 and 116 from an internal memory or from a remote (e.g., cloud-hosted) memory service. In some cases, the computer 122 can look for the 3D shapes 114 and 116 in a library of 3D shapes by searching for a particular protein in the protein-protein complex 104, or by searching for one or more homologs of one or both of the proteins, or by combining different portions of such homologs to create new homologs.

A number of docking models are accessed (308), each of which defines a candidate pose pair. For example, the computer 122 may access the models 110 from an internal memory or from a remote (e.g., cloud-hosted) memory service. In some cases, the computer 122 may search all possible models 110 available. In some cases, the computer 112 may search a subset of all possible models 110 by querying only for models with certain parameters specified based on the technical requirements of the process 300.

For each docking model, the first 3D shape, the second 3D shape, and the candidate pose pair are applied (310), and a corresponding fit score is generated for the docking model that describes how well the pose pair fits the docking model. For example, computer 122 can calculate the fit score for the single model 110 by feeding the composite image 108, the image mask 112, the 3D shapes 114 and 116, and the single model 110 to a fit function, which performs calculations on the inputs and returns a numerical value indicating how well the model describes a particular state of the other input data. Computer 122 can repeat this for each model 110.

Based on the fit scores, one of the docking models is selected as the detected model for the protein-protein complex (312). For example, the computer 122 can select the best model 110 based on the fit scores of each model 110 and possibly other data, as described below.

FIG. 4 illustrates an example process 400 for generating a composite image, e.g., as part of pre-processing performed prior to accessing 302 the composite image. For example, process 400 can be performed in conjunction with elements of system 100, and for clarity, the examples herein are described in terms of elements of system 100. However, other systems may be used to perform process 400 and other similar processes.

The protein-protein complex sample is loaded into a cryo-electron microscope (402). For example, a human operator and/or an automated service machine (e.g., a material handling robot) can cryo-cool the protein-protein complexes 104 and embed them in a medium such as vitreous water. The solution can be applied to a grid mesh and frozen in a cooling medium such as liquid ethane. The mesh can then be loaded into the imager 102.

A number of composite images are generated (404). For example, a human operator, an automated service machine, and/or a computer 122 can instruct the imager 102 to perform electron microscopy on the composite 104 to generate a composite image 106. Once generated, the composite image 106 can be stored in computer memory (e.g., internal to the computer 122 or at an external location).

A composite image may be generated (406) from the multiple composite images. For example, for each pixel location, computer 122 may aggregate the pixel values by calculating an average and storing the average at a given pixel location across multiple pixels of image 106 to create a single aggregate pixel value, which may be stored in composite image 108. In some cases, this aggregation may be a weighted average, may exclude outliers, may include a median or mode, etc.

In some cases, generating a composite image involves extracting protein-protein-complex sub-images from the complex image; and sorting and orienting the sub-images. For example, computer 122 can examine each of images 106 and find pixel areas that show the complex, and copy these pixel values into separate sub-image files. In another example, computer 122 can do this without using separate files, but for clarity, separate files are described. Then, for each separate file, the computer can modify the sub-images so that each sub-image shows the protein with the same orientation, scale, intensity, etc. As will be appreciated, this can include one or more image manipulation processes.

FIG. 5 illustrates an example process 500 for masking a composite image, e.g., as part of masking composite image 304. For example, process 500 can be performed in conjunction with elements of system 100, and for clarity, the examples herein are described in terms of elements of system 100. However, other systems may be used to perform process 500 and other similar processes.

A masking graphic user interface (GUI) is presented to the user (502). For example, the computer 112 can load the GUI, such as an application interface or a web page on a screen. The screen can render an image of the composite image 108 along with interface elements (e.g., buttons, scroll bars) that receive user input. User input can be provided by a human operator pressing physical buttons, moving a pointing device, tapping a touch screen, etc.

A first user input is received (504) specifying the unmasked portion. For example, the user may use an interface element to specify multiple (e.g., 3, 4, 6, or 9) points on the rendered composite image 108. For example, the user may use domain knowledge to visually identify areas of the composite image 108 that are likely to represent docking areas of the antigen-specified protein-protein complex 104. The user may then use a pointing device, such as a mouse, to identify four vertices of a bounding box that is drawn around the identified area.

The bounding box is generated by connecting the locations specified by the first user input (506). For example, the computer 122 can computationally generate line segments terminating at successive points identified by the user, including line segments terminating at the first location and the last location. This can generate a fully connected polygon.

The portion outside the box is marked as the masked portion (508) and the portion inside the box is marked as the unmasked portion (510). For example, each pixel that is fully or partially inside the polygon can be given a color value in the image mask 112 (e.g., black, white) and each pixel that is fully or partially outside the polygon can be given a different color value (e.g., white, black).

FIG. 6 illustrates an example process 600 for selecting a detected model from a group of candidate models, e.g., as part of selecting detected model 312. For example, process 600 can be performed in conjunction with elements of system 100, and for clarity, the examples herein are described in terms of elements of system 100. However, other systems may be used to perform process 600 and other similar processes.

The candidate docking model with the best fit score is selected (602). For example, when data is applied to the models 310, a fit score is calculated for each model. The fit score can be viewed as a measure of how well the model 110 predicts the arrangement of colors in the composite image 108 given the image mask 112 and the 3D shapes 114 and 116. In some cases, the fit score is generated by projecting various orientations of the docking model into the 2D image and comparing the projected images to the sensed composite image 106. The projection that produces the smallest difference to the sensed composite is scored with the best cross-correlation score.

The fit scores for each model 110 are used to identify a subset of models 110 with the best fit scores. In some cases, these are the top scoring models 110. These can be found by the computer 122 selecting the docking models 110 with the N (e.g., 5, 10, 20, 100) highest fit scores. In some cases, these are any models that are sufficiently predictive. These can be found by the computer 122 selecting all docking models 110 with fit scores above a threshold M (e.g., 0.8, 0.9, 0.095, 0.0999 on a scale of 0 to 1).

The candidate models are presented in a user interface (604) and a user selection input is received (606) to select one of the subset of docking models as the detected model. For example, the computer 122 can display the subset of models 110 by rendering each docking model and showing the associated score 118 along with a rendering of the composite image 108. The user can select one using an input device. In some cases, all candidate docking models 110 are shown simultaneously, allowing the user to review all options simultaneously, allowing for more convenient and accurate consideration.

7 illustrates an example of a computing device 700 and an example of a mobile computing device that can be used to implement the techniques described herein. The computing device 700 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. The mobile computing device is intended to represent various forms of mobile devices, such as personal digital assistants, mobile phones, smartphones, and other similar computing devices. The components shown, their connections and relationships, and their functions are intended to be illustrative only and are not intended to limit the implementation of the invention described and/or claimed herein.

The computing device 700 includes a processor 702, a memory 704, a storage device 706, a high-speed interface 708 that connects to the memory 704 and multiple high-speed expansion ports 710, and a low-speed interface 712 that connects to the low-speed expansion port 714 and the storage device 706. Each of the processor 702, memory 704, storage device 706, high-speed interface 708, high-speed expansion port 710, and low-speed interface 712 are interconnected using various buses and may be mounted on a common motherboard or in other manners as appropriate. The processor 702 may process instructions for execution within the computing device 700, including instructions stored in the memory 704 and/or the storage device 706, to display graphical information for a GUI on an external input/output device, such as a display 716 coupled to the high-speed interface 708. In other implementations, multiple processors and/or multiple buses may be used as appropriate, along with multiple memories and memory types. Also, multiple computing devices may be connected, with each device providing a portion of the required operations (e.g., as a server bank, a blade server group, or a multiprocessor system).

Memory 704 stores information within computing device 700. In some implementations, memory 704 is a volatile memory unit. In some implementations, memory 704 is a non-volatile memory unit. Memory 704 may also take another form of computer-readable medium, such as a magnetic disk or optical disk.

The storage device 706 can provide mass storage for the computing device 700. In some embodiments, the storage device 706 can be or include a floppy disk device, a hard disk device, an optical disk device, or a computer-readable medium such as a tape device, a flash memory or other similar solid-state memory device, or an array of devices including devices in a storage area network or other configuration. The computer program product can be tangibly embodied in an information carrier. The computer program product can also include instructions that, when executed, perform one or more methods, such as the methods described above. The computer program product can also be tangibly embodied on a computer or machine-readable medium, such as memory 704, the storage device 706, or memory on the processor 702.

The high-speed interface 708 manages bandwidth-intensive operations for the computing device 700, while the low-speed interface 712 manages less bandwidth-intensive operations. Such allocation of functions is merely exemplary. In some embodiments, the high-speed interface 708 is coupled to the memory 704, the display 716 (e.g., through a graphics processor or accelerator), and a high-speed expansion port 710 that can receive various expansion cards (not shown). In an embodiment, the low-speed interface 712 is coupled to the storage device 706 and the low-speed expansion port 714. The low-speed expansion port 714, which can include various communication ports (e.g., USB, Bluetooth, Ethernet, wireless Ethernet), can be coupled to one or more input/output devices, such as a keyboard, a pointing device, a scanner, or a networking device such as a switch or router, for example, through a network adapter.

As shown, computing device 700 can be implemented in a number of different forms. For example, computing device 700 can be implemented multiple times as a standard server 720 or in a group of such servers. In addition, computing device 700 can be implemented in a personal computer, such as a laptop computer 722. Computing device 700 can also be implemented as part of a rack server system 724. Alternatively, components from computing device 700 can be combined with other components in a mobile device (not shown), such as mobile computing device 750. Each such device can include one or more of computing device 700 and mobile computing device 750, and the overall system can include multiple computing devices in communication with each other.

The mobile computing device 750 includes, among other components, a processor 752, a memory 764, an input/output device such as a display 754, a communication interface 766, and a transceiver 768. The mobile computing device 750 may also be provided with a storage device such as a microdrive or other device to provide additional storage. Each of the processor 752, memory 764, display 754, communication interface 766, and transceiver 768 are interconnected using various buses, and some of the components may be mounted on a common motherboard or in other manners as appropriate.

The processor 752 can execute instructions within the mobile computing device 750, including instructions stored in the memory 764. The processor 752 can be implemented as a chipset of chips including separate and multiple analog and digital processors. The processor 752 can provide for the coordination of other components of the mobile computing device 750, such as control of a user interface, applications executed by the mobile computing device 750, and wireless communication by the mobile computing device 750.

The processor 752 can communicate with a user through a control interface 758 and a display interface 756 coupled to a display 754. The display 754 can be, for example, a TFT (thin film transistor liquid crystal display) display or an OLED (organic light emitting diode) display, or other suitable display technology. The display interface 756 can include suitable circuitry for driving the display 754 to present graphical and other information to the user. The control interface 758 can receive commands from the user and translate those commands for sending to the processor 752. In addition, an external interface 762 can provide communication with the processor 752 to enable near area communication of the mobile computing device 750 with other devices. The external interface 762 can provide, for example, wired communication in some implementations or wireless communication in other implementations, and multiple interfaces can also be used.

The memory 764 stores information within the mobile computing device 750. The memory 764 may be embodied as one or more of a computer readable medium, a volatile memory unit, or a non-volatile memory unit. An expansion memory 774 may also be provided and connected to the mobile computing device 750 through an expansion interface 772, which may include, for example, a SIMM (single in-line memory module) card interface. The expansion memory 774 may provide additional storage space for the mobile computing device 750 or may store applications or other information related to the mobile computing device 750. In particular, the expansion memory 774 may include instructions that perform or supplement the processes described above, and may also include secure information. Thus, for example, the expansion memory 774 may be provided as a security module of the mobile computing device 750 and may be programmed with instructions that enable secure use of the mobile computing device 750. In addition, secure applications may be provided by the SIMM card along with additional information, such as placing identifying information on the SIMM card in a manner that cannot be hacked.

The memory may include, for example, flash memory and/or NVRAM memory (non-volatile random access memory), as discussed below. In some embodiments, the computer program product is tangibly embodied in an information carrier. The computer program product includes instructions that, when executed, perform one or more methods, such as the methods described above. The computer program product may be a computer or machine readable medium, such as memory 764, expansion memory 774, or memory on processor 752. In some embodiments, the computer program product may be received in a propagated signal, for example, via transceiver 768 or external interface 762.

The mobile computing device 750 may communicate wirelessly through a communication interface 766, which may include digital signal processing circuitry as required. The communication interface 766 may provide communications under various modes or protocols, such as GSM voice telephony (Global System for Mobile Communications), SMS (Short Message Service), EMS (Enhanced Messaging Service), or MMS messaging (Multimedia Messaging Service), CDMA (Code Division Multiple Access), TDMA (Time Division Multiple Access), PDC (Personal Digital Cellular), WCDMA (Wideband Code Division Multiple Access), CDMA2000, or GPRS (General Packet Radio Service), among others. Such communications may be through a transceiver 768 using radio frequencies, for example. In addition, short-range communications may be performed, such as using Bluetooth, WiFi, or other such transceivers (not shown). In addition, a GPS (Global Positioning System) receiver module 770 may provide the mobile computing device 750 with additional navigation- and location-related radio data that may be used by applications executing on the mobile computing device 750, as appropriate.

The mobile computing device 750 can also provide voice communication using a voice codec 760 that can receive spoken information from a user and convert that information into usable digital information. Similarly, the voice codec 760 can generate audible sound for the user, such as through a speaker in a handset on the mobile computing device 750. Such sound can include sound from a voice telephone call, can include recorded sound (e.g., voice messages, music files, etc.), and can also include sound generated by applications running on the mobile computing device 750.

As shown, the mobile computing device 750 may be implemented in a number of different forms. For example, the mobile computing device 750 may be implemented as a mobile phone 780. The mobile computing device 750 may also be implemented as part of a smartphone 782, a personal digital assistant, or other similar mobile device.

Various implementations of the systems and techniques described herein may be realized in digital electronic circuitry, integrated circuits, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various implementations may include implementation in one or more computer programs executable and/or interpretable on a programmable system including at least one programmable processor, which may be special purpose or general purpose, coupled to receive data and instructions from and transmit data and instructions to a storage system, at least one input device, and at least one output device.

These computer programs (also known as programs, software, software applications, or code) contain machine instructions for a programmable processor and may be implemented in high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages. As used herein, the terms machine-readable medium, computer-readable media refer to any computer program product, apparatus, and/or device (e.g., magnetic disk, optical disk, memory, programmable logic device (PLD)) used to provide machine instructions and/or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term machine-readable signal refers to any signal used to provide machine instructions and/or data to a programmable processor.

To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user, as well as a keyboard and pointing device (e.g., a mouse or trackball) by which the user can provide input to the computer. Other types of devices can also be used to provide interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or haptic feedback); input from the user can be received in any form, including acoustic, speech, or tactile input.

The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or includes middleware components (e.g., an application server), or includes front-end components (e.g., a client computer having a graphical user interface or web browser through which a user can interact with an implementation of the systems and techniques described herein), or includes any combination of such back-end, middleware, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communications network). Examples of communications networks include a local area network (LAN), a wide area network (WAN), and the Internet.

A computing system may include clients and servers. Clients and servers are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

Claims (32)

1. A method for detecting a protein-protein complex interaction, comprising:
accessing a composite image comprising a plurality of complex images of a sample of a protein-protein complex comprising a first protein and a second protein;
masking the composite image to generate a masked portion and an unmasked portion;
accessing a first three-dimensional (3D) shape of a first protein and a second 3D shape of a second protein;
accessing a plurality of docking models, each of which defines a candidate pose pair;
applying, for each docking model, the first 3D shape, the second 3D shape, and the candidate pose pairs to generate, for the docking model, a corresponding fit score describing how well the pose pairs fit the docking model;
selecting one of the docking models as a detected model for the protein-protein complex based on the fitness score;
The method comprising: The method of claim 1, further comprising generating a plurality of composite images. The method of claim 2, further comprising generating a composite image from a plurality of complex images, where generating the composite image comprises extracting protein-protein-complex sub-images from the complex images; and orienting and classifying the sub-images. The method of any one of claims 1 to 3, wherein the composite image is a cryo-electron microscopy (cryo-EM) image. The method of claim 1, wherein each complex image includes a plurality of pixels, each having an address and carrying a color value that represents a corresponding portion of the protein-protein complex sample. The method of any one of claims 1 to 3, wherein the composite image includes a plurality of pixels, each having an address and holding a color value that is a collection of color values of pixels having the same address in each of the plurality of composite images. The method of any one of claims 1 to 3, wherein the composite image is masked without specific user input. The method of claim 1, wherein the composite image is masked, and the masking includes receiving a first user input specifying a non-masked portion. Receiving a first user input specifying an unmasking portion includes:
generating a bounding box by connecting locations specified by a first user input;
recording a portion of the composite image within the bounding box as an unmasked portion;
The method of claim 8 , comprising: The method of any one of claims 1 to 3, wherein the first 3D shape is indexed as a first protein and the second 3D shape is indexed as a second protein. The method of claim 1, wherein the first 3D shape is indexed as a first homolog of the first protein. The method of claim 11, wherein the second 3D shape is indexed as a second homolog of the second protein. The method of any one of claims 1 to 3, wherein the candidate pose pairs include a candidate location, a candidate orientation, and a candidate docking area. The method of any one of claims 1 to 3, wherein the compatibility score is a cross-correlation score generated by projecting the docking model onto the 2D model. Selecting one of the docking models as a detected model for the protein-protein complex based on the fitness score includes:
A subset of docking models:
i) selecting docking models with top N fitness scores;
ii) selecting all docking models having a fitness score above a threshold M;
2. The method of claim 1, further comprising: discriminating between the first and second objects based on their corresponding match scores by one of the group consisting of: The method of claim 15, wherein selecting one of the docking models as the detected model for the protein-protein complex includes receiving a second user input that selects one of the subset of docking models as the detected model. 1. A system for detecting protein-protein complex interactions comprising:
one or more processors;
a computer memory for storing instructions;
the instructions, when executed by a processor, cause the processor to:
accessing a composite image comprising a plurality of complex images of a sample of a protein-protein complex comprising a first protein and a second protein;
accessing a first three-dimensional (3D) shape of a first protein and a second 3D shape of a second protein;
accessing a plurality of docking models, each of which defines a candidate pose pair;
applying, for each docking model, the first 3D shape, the second 3D shape, and the candidate pose pair to generate, for the docking model, a corresponding fit score describing how well the pose pair fits the docking model;
selecting one of the docking models as a detected model for the protein-protein complex based on the fitness score;
The system performs the following operations: The system of claim 17, wherein the operation further includes generating a plurality of composite images. The system of claim 18, wherein the operations further include generating a composite image from a plurality of complex images, where generating the composite image includes extracting protein-protein-complex sub-images from the complex images; and orienting and classifying the sub-images. The system of any one of claims 17 to 19, wherein the composite image is a cryo-electron microscopy (cryo-EM) image. The system of claim 17, wherein each complex image includes a plurality of pixels, each having an address and carrying a color value that represents a corresponding portion of the protein-protein complex sample. 22. The system of claim 21, wherein the composite image includes a plurality of pixels, each having an address and holding a color value that is a collection of color values of pixels having the same address in each of the plurality of composite images. The system of any one of claims 17 to 19, wherein the composite image is masked without specific user input. The system of claim 17, wherein the composite image is masked, and the masking includes receiving a first user input designating an unmasked portion. Receiving a first user input specifying an unmasking portion includes:
generating a bounding box by connecting locations specified by a first user input;
recording a portion of the composite image within the bounding box as an unmasked portion;
25. The system of claim 24, comprising: The system of any one of claims 17 to 19, wherein the first 3D shape is indexed as a first protein and the second 3D shape is indexed as a second protein. The system of any one of claims 17 to 19, wherein the first 3D shape is indexed as a first homolog of the first protein. 20. The method of any one of claims 17 to 19, wherein the second 3D shape is indexed as a second homolog of the second protein. The system of any one of claims 17 to 19, wherein the candidate pose pairs include a candidate location, a candidate orientation, and a candidate docking area. The system of any one of claims 17 to 19, wherein the compatibility score is a cross-correlation score generated by projecting the docking model onto the 2D image. Selecting one of the docking models as a detected model for the protein-protein complex based on the fitness score includes:
A subset of docking models:
i) selecting docking models with top N fitness scores;
ii) selecting all docking models having a fitness score above a threshold M;
20. The system of claim 17, further comprising: discriminating between the first and second objects based on their corresponding match scores by one of the group consisting of: The system of claim 31, wherein selecting one of the docking models as the detected model for the protein-protein complex includes receiving a second user input that selects one of the subset of docking models as the detected model.

Images