TomoNet: A streamlined cryoET software pipeline with automatic particle picking on flexible lattices

TomoNet: A streamlined cryoET software pipeline with automatic particle picking on flexible lattices | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results TomoNet: A streamlined cryoET software pipeline with automatic particle picking on flexible lattices View ORCID Profile Hui Wang , View ORCID Profile Shiqing Liao , View ORCID Profile Xinye Yu , View ORCID Profile Jiayan Zhang , View ORCID Profile Z. Hong Zhou doi: https://doi.org/10.1101/2024.02.17.580557 Hui Wang 1 Department of Bioengineering, University of California , Los Angeles (UCLA), Los Angeles, CA 90095, USA 2 California NanoSystems Institute , UCLA, Los Angeles, CA 90095, USA 3 Department of Microbiology, Immunology, and Molecular Genetics, UCLA, Los Angeles , CA 90095, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Hui Wang For correspondence: Hong.Zhou{at}UCLA.edu Shiqing Liao 2 California NanoSystems Institute , UCLA, Los Angeles, CA 90095, USA 3 Department of Microbiology, Immunology, and Molecular Genetics, UCLA, Los Angeles , CA 90095, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Shiqing Liao Xinye Yu 3 Department of Microbiology, Immunology, and Molecular Genetics, UCLA, Los Angeles , CA 90095, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Xinye Yu Jiayan Zhang 2 California NanoSystems Institute , UCLA, Los Angeles, CA 90095, USA 3 Department of Microbiology, Immunology, and Molecular Genetics, UCLA, Los Angeles , CA 90095, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Jiayan Zhang Z. Hong Zhou 1 Department of Bioengineering, University of California , Los Angeles (UCLA), Los Angeles, CA 90095, USA 2 California NanoSystems Institute , UCLA, Los Angeles, CA 90095, USA 3 Department of Microbiology, Immunology, and Molecular Genetics, UCLA, Los Angeles , CA 90095, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Z. Hong Zhou For correspondence: Hong.Zhou{at}UCLA.edu Abstract Full Text Info/History Metrics Supplementary material Data/Code Preview PDF ABSTRACT Cryogenic electron tomography (cryoET) is capable of determining in situ biological structures of molecular complexes at near atomic resolution by averaging half a million subtomograms. While abundant complexes/particles are often clustered in arrays, precisely locating and seamlessly averaging such particles across many tomograms present major challenges. Here, we developed TomoNet, a software package with a modern graphical user interface to carry out the entire pipeline of cryoET and subtomogram averaging to achieve high resolution. TomoNet features built-in automatic particle picking and 3D classification functions and integrates commonly used packages to streamline high-resolution subtomogram averaging for structures in one-, two- or three-dimensional arrays. Automatic particle picking is accomplished in two complementary ways: one based on template matching and the other employing deep learning. TomoNet’s hierarchical file organization and visual display facilitate efficient data management as required for large cryoET datasets. Applications of TomoNet to three types of datasets demonstrate its capability of efficient and accurate particle picking on flexible and imperfect lattices to obtain high-resolution 3D biological structures: virus-like particles, bacterial surface layers within cellular lamellae, and membranes decorated with nuclear egress protein complexes. These results demonstrate TomoNet’s potential for broad applications to various cryoET projects targeting high-resolution in situ structures. INTRODUCTION Single-particle cryogenic electron microscopy (cryoEM) is employed to elucidate atomic-level structures of purified biological complexes. This methodology adheres to a standardized and well-established workflow supported by advanced software packages such as Relion 1 and cryoSparc 2 . In parallel, cryogenic electron tomography (cryoET), coupled with subtomogram averaging (STA), expands the investigative scope to encompass heterogeneous macromolecules in their native context 3 – 10 . To enhance the resolution of subunits within in situ macromolecules, subtomograms ( i.e., particles) are extracted from each tomogram and then subjected to 3D alignment and averaging, thereby improving signal-to-noise ratio. Notably, STA has achieved resolutions up to sub-3 Å for in situ structures of large cellular complexes such as ribosomes, approaching the capabilities of single-particle cryoEM methodologies 11 – 14 . The workflow for cryoET and STA typically involves five key components across specific software packages. In cryoET preprocessing, dose fractionated frames are collected from an electron microscope, undergo motion correction, organized, and then assembled into individual tilt series. In tomogram reconstruction, three-dimensional reconstructions are generated from those tilt series. In particle picking, particles of interest are identified and extracted from tomograms. Complexity varies based on the diverse and intricate nature of in situ cellular samples and their unique configurations. Many packages include their own particle picking methods, such as oversampling using a supporting geometry in Dynamo 15 , 16 , template matching in emClarity 17 and machine learning in crYOLO 18 . In 3D refinement and classification, particles are iteratively classified and refined to obtain a final structure at sub-nanometer or near atomic resolution, which has been demonstrated by software packages like Relion 13 , 19 , emClarity 17 , EMAN2 4 and Warp 20 . Finally, activities in post-processing include map sharpening, Fourier shell correlation (FSC) calculation, visualization by placing averaged maps back into the original tomogram, etc. Users often need to navigate between several specialized software packages for optimal results, which often demands a certain level of computational proficiency that poses a barrier for many. The method for particle picking varies on a case-by-case basis, dictated by the characteristics of in situ cellular samples. In the early works of STA, manual particle picking was employed, particularly when aiming for resolutions between 20-50 Å with a maximum of several hundred particles 21 – 23 . However, for biological samples exhibiting periodic structures, oversampling on specified geometry was leveraged to significantly reduce the labor associated with acquiring enough particles for improved resolutions. For instance, HIV virus-like particles (VLPs) adopt a hexagonal Gag protein lattice in its sphere-like configuration 16 . Other examples include the Marburg Virus 24 , Herpes simplex virus 25 , and the Coat protein complex II 26 , all of which contain lattice-like arrangements with repeating subunits that could benefit from particle picking automation when performing cryoET data processing. With an increasing demand for automation to enhance efficiency with minimal manual intervention, template matching has emerged as a popular method for automatic particle picking, relying on a user-provided reference map 17 , 27 . Simultaneously, convolutional neural networks have shown promising results for cryoET automatic particle picking given its capacity to analyze three-dimensional feature maps and autonomously identify prominent features within specific samples 28 – 31 . These machine learning approach typically operate template-free and often obviates the need for human annotation 32 . The expanding array of specialized software tools designed for specific tasks posts a critical need for seamless software integration within the cryoET workflow. Transitioning between various software packages can be a cumbersome process. Remarkably, recent initiatives have made notable progress in tackling this integration challenge. For example, TomoBEAR 33 offers an integrated solution, while ScipionTomo 34 and nextPYP 35 provide a comprehensive web-based platform for managing various tasks in the cryoET pipeline. Notably, none of these packages takes specific advantage of the fact that abundant complexes exist in arrays of some sort, albeit with imperfections, variability, or flexibility. In this context, we have developed TomoNet, a software package designed for streamlining the cryoET and STA data processing workflow, with a modern GUI ( Figure 1 and Figure 2 ). Our methodology employs a geometric template matching approach, rooted in the concept of “Auto Expansion”, which serves as a general particle picking solution for biological complexes organized in flexible, variable, or imperfect arrays. TomoNet is also powered by a deep learning-based solution to automate particle picking, which only needs 1-3 tomograms with known particle locations as ground truth for model training. Importantly, while TomoNet is particularly powerful for locating and averaging particles arranged on flexible or imperfect lattices, it can be applied to a broader range of particle types, offering a more generalizable trained model. These methods significantly diminish the need for manual inputs, and their outcomes can be seamlessly imported into Relion for subsequent high-resolution 3D classifications and refinements. We demonstrate the capabilities of TomoNet by applying it to three datasets with distinct protein lattice types, highlighting its accuracy and efficiency in identifying particles across diverse scenarios. Download figure Open in new tab Figure 1, Illustration of TomoNet’s comprehensive pipeline for cryoET and STA. The pink border encloses the sequential functions implemented in TomoNet, and they can be subdivided into three principal segments, delineated by the orange borders. These segments include tomogram preparation on the left, template matching-based particle picking “Auto Expansion” in the center, and deep learning-based automatic particle picking on the right. Download figure Open in new tab Figure 2, A screenshot of TomoNet GUI. The TomoNet GUI contains three main areas: the menu bar (top left), the input and operate area (top right), and the log window (bottom). Bottom left: results generated by the “3D Subtomogram Place Back” function can be visualized in ChimeraX. Bottom right: intermediate results of picked particles viewed with IMOD. RESULTS Overall design of TomoNet TomoNet is a Python-based software package that integrates commonly used cryoET packages to streamline the cryoET and STA pipeline, with a particular emphasis on automating particle picking of lattice-configured structures and cryoET project management. As shown in the main menu and the entire TomoNet pipeline ( Figure 1 and Figure 2 ), after data collection from electron microscopy, TomoNet can perform motion correction with integration of MotionCorr2 36 ; tilt series assembly and tomogram reconstruction with integration of IMOD 37 and AreTomo 38 ; CTF estimation with integration of CTFFIND4 39 ; manual particle picking with IMOD; particle picking using built-in geometric template matching-based algorithms with integration of PEET 40 ; automatic particle picking using built-in deep learning-based algorithms; 3D classification/particle cleaning and subtomograms placing back with built-in algorithms. This design also allows on-the-fly tomogram reconstruction processing during data collection, which facilitates a quick quality check. TomoNet generates particle picking results in STAR format 41 , which can be incorporated into Relion for high-resolution 3D refinement. It can also read Relion results in STAR format for particle cleaning and subtomograms placing back ( Figure 1 ). Particle picking with “Auto Expansion” The “Auto Expansion” module is based on template matching and uses cross-correlation coefficient as a selection criterion, with a design to pick particles on flexible lattices with minimal manual inputs, its basic concept is elucidated in Figure 3 . These particles exist in array-like configurations and manifest as flexible, partial, and imperfect lattices in one, two and three dimensions (1-3D). Examples are abound: microtubule doublets, ubiquitous in most cells, consist of 96 nm axonemal 1D translational repeat units 22 , 42 (1D rotational lattice); HIV VLPs 43 and surface layer (S-layer) lattice of prokaryotic cells 44 , 45 are composed of hexametric subunits (2D lattice); paraflagellar rod of protozoan species is organized into para-crystalline arrays in its distal zone 21 (3D lattice). In TomoNet, each of these isolated lattice densities is called a patch, within which all subunits of the complex are connected. For instance, Figure 3 illustrates two patches with different sizes. Download figure Open in new tab Figure 3, Illustration of the first two iterations of “Auto Expansion“ particle picking. There are two patches of a hexagonal lattice with individual particles represented by solid hexagons. At iteration 0, 18 “candidate” particles (dashed blue) were selected from the neighbors of 3 “seed” particles (orange). 14 good particles remained and will serve as “seed” particles in iteration 1, and 3 “seed” particles in iteration 0 were saved in the final particle set (green). At iteration 1, 35 “candidate” particles were selected from the neighbors of 14 “seed” particles. 29 good particles remained and will serve as “seed” particles in iteration 2, and 14 “seed” particles were saved in the final particle set. “Auto Expansion” is an iterative process and will stop when no “candidate” can be detected. “Auto Expansion” is an iterative process; each iteration expands the particle set by adding more unpicked ones. To initiate “Auto Expansion”, users need to prepare a few “seed” particles that sparsely distribute across all observed patches. Typically, the numbers of such “seed” particles per tomogram range from 20 to 200, which depends on the number and size of patches in the input tomogram. Then, “Auto Expansion” iteratively expands the “seed” particle set to a final particle set that contains all particles on given flexible lattices, following three steps for each iteration ( Figure 3 ). Firstly, potential particles adjacent to each “seed” particle are calculated and selected as “candidate” particles. Secondly, these “candidate” particles undergo alignments to a user-provided reference and are evaluated based on cross-correlation coefficient, such that “wrong” particles with low cross-correlations are excluded. Thirdly, qualified “candidate” particles are added to the particle set and become “seed” particles for the next iteration. During this process, only unpicked ones can be considered as “candidate” particles, and “Auto Expansion” stops either when no “candidate” particles are detected or when the user-defined maximum iteration number is reached. Doing this allows for an exhaustive exploration of particles on given lattices following their assembly topology with no restriction on geometry and outputs a final particle picking result ( Figure 2 ). Compared with conventional template matching methods, “Auto Expansion” incorporates prior knowledge of lattice configuration to iteratively guide the search for “candidate” particles, i.e. , unpicked particles following user-defined paths, as detailed in the Method section and TomoNet’s user manual. Thus, “Auto Expansion” significantly reduces computational complexity by searching in the regions of interest only, with restricted angular and translational search ranges defined by users. As a result, it reduces the number of incorrectly picked particles. Notably, “Auto Expansion” potentially works for any flexible, imperfect, or variable lattices in 1D, 2D and 3D and has no intrinsic size limit of subunits. Automatic particle picking by deep learning The “AI AutoPicking” module is designed for automatic particle picking using supervised machine learning, which employs a U-net convolutional neural network for model training. There are three main steps in “AI AutoPicking”: training data preparation, neural network training, and particle coordinate prediction, as detailed in the Method section ( Figure 4 ). It only requires an input training dataset consisting of 1-3 tomograms paired with their corresponding particles coordinate files. The trained model can then be applied on the entire tomography dataset and output predicted particles for each tomogram. Download figure Open in new tab Figure 4, Illustration of “AI AutoPicking“ process consisting of three steps. The HIV dataset was used for this illustration, and the particles refer to Gag hexamers. a, Training dataset preparation. Using the user-provided tomograms with associated particle coordinate files, subtomograms containing particle densities were extracted. For each subtomogram, TomoNet generated a segmentation map based on the coordinates of particles, where the voxels near a particle’s center are shown as white and the others as black. b, Neural network training. The generated subtomograms and segmentation maps were used as the input and output to train the convolutional neural network in learning how to segment out particle densities. c, Particle coordinate prediction. Firstly, TomoNet applied the trained neural network model to unseen tomograms and generated associated predicted segmentation maps. Then, the particle coordinate information was obtained from the segmentation maps using clustering algorithms. Essentially, the neural network in “AI AutoPicking” is trained as a voxel-wise binary classifier, which determines whether a voxel in density maps is part of a particle ( Figure 4b ). To prepare for training, data pairs (ground truth) consist of extracted subtomograms coupled with their associated segmentation maps, within where each particle is labeled by a cube near its center ( Figure 4a ). The trained neural network model can be applied on other tomograms to perform particle segmentation. Finally, the particles coordinate information can be retrieved from the predicted segmentation maps ( Figure 4c ). 3D classification using TomoNet In addition to the above two commentary modules for particle picking, TomoNet allows users to eliminate “bad” particles based on user-defined geometric constraints, which could serve as 3D classification during high-resolution particle refinements. Lattice variation in cryoET data has multiple plausible causes. Biologically, particles may be incomplete near the lattice edge due to paused biology assembly process 46 . Experimentally, lattices tend to become flattened near the air-water interface of the sample during imaging. These variabilities pose challenges for 3D classification in the process of high-resolution STA, making it difficult to exclude “bad” particles that exhibit unexpected coordinates and orientations assignment as subunits of lattices (Supplementary Movie 1). Removing these “bad” particles is necessary for achieving better resolutions 47 . To accomplish this, TomoNet assesses each particle by counting its neighboring particles and calculating the averaged tilt angle to these neighbors to represent local surface curvature of a lattice. TomoNet identifies particles with too few neighbors or large tilt angles to their neighbors as “bad” particles since they potentially deviate from the lattice configuration. This step can be integrated into high-resolution refinement in Relion, providing an alternative 3D classification method based on analyzing spatial relationships between particles. Application to in situ viral protein arrays: the matrix protein lattice in HIV VLPs To validate TomoNet as an integrated high-resolution cryoET and STA pipeline and an efficient particle picking tool, four tomograms were processed from the HIV-1 Gag dataset which resolved the Gag hexamer structure at 3.2 Å resolution. Motion corrected images underwent tilt series assembly, CTF estimation, and tomographic reconstruction using TomoNet. Within these tomograms, the VLP hexagonal lattice and its building blocks were observed, and some of these observed VLPs exhibited sphere-like geometry ( Figure 5a ). Download figure Open in new tab Figure 5, TomoNet application to arrays of matrix protein in HIV VLPs. a, Illustration of picked “seed” particles on a spherical VLP. Green segments represent the particles’ Y-axis. Scale bar is 20 nm. b, “Auto Expansion” result on three VLPs within tomogram TS_01, with yellow dots representing the center of the hexamer subunits. c, “AI AutoPicking” particle prediction result of tomogram TS_45 shows its ability to pick particles on all lattices of different sizes and shapes. d, Visualization of three different variations of the HIV Gag lattices generated by placing back averaged structures, two exhibiting a spherical shape, and one presented as a fragment. Blue arrows indicate defects in the lattice. As detailed in the Method section, a combination of “Auto Expansion” and “AI AutoPicking” was applied on the above four tomograms; as a result, particles were readily picked on all the observed lattice patches ( Figure 5b, c ). Then, these picked particles were imported to Relion to perform high-resolution particle refinements, resulting in a final reconstruction of the Gag hexamer structure ( Figure 6 ). Download figure Open in new tab Figure 6, Final map resolution of HIV Gag hexamer. a, Final reconstruction of Gag hexamer (grey) fitted with the atomic model (PDB: 5l93). b, One segmented Gag monomer structure, inset shows a closer view of carboxy-terminal domain overlay with the atomic model. c, Directional Fourier shell correlation (FSC) curves for the STA of Gag hexamer structure, with a global resolution at 3.2 Å. Using the “3D subtomogram place back” function in TomoNet, 3D visualizations were generated to illustrate the in situ assembly of the VLP lattices ( Figure 5d and Figure 7 ). All VLP lattices with various sizes and shapes were captured even with irregular shapes ( Figure 7e and Supplementary Movie 2), demonstrated TomoNet’s particle picking ability on flexible lattices. Lattice defects on each VLP were also identified consistent with previous studies 48 , enhancing the understanding of lattice assembly mechanisms 49 . Download figure Open in new tab Figure 7, Comparative visualization of lattices obtained from TomoNet and Relion tutorial. a, b, Visualized comparison of particles used in TomoNet and Relion tutorial within tomogram TS_01. TomoNet can pick particles not only on a sphere-like lattice but also on others with random shapes. c, d, A comparison of particle picking results on two sphere-like shape VLPs from TomoNet and Relion tutorial. e, A zoom-in view of an irregularly shaped lattice. Coloring is based on surface curvatures at the point of each subunit. Application to focused ion beam (FIB)-milled cellular sample: the S-layer lattice of prokaryotic cell We validated TomoNet’s particle picking capability by processing one tomogram of FIB-milled Caulobacter crescentus cells from EMD-23622 50 . The S-layer functions as a component of the cell wall covering the cell body. Thus, its lattice geometry is typically defined by the shape of cells ( Figure 8a ). The pleomorphic shape of C. crescentus cell in variable sizes, with the low contrast shown in this tomogram, hindered locating subunits on the S-layer lattice and raised difficulty for efficient particle picking on its S-layer lattice ( Figure 8a ). Download figure Open in new tab Figure 8, TomoNet application to S-layer structure in FIB-milled cellular sample. a, A tomographic slice view shows two C. crescentus cells in a FIB-milled lamella. b, Orthogonal slice views of the averaged density map generated in TomoNet, showing the hexagonal distribution of S-layer inner domains. Scale bar is 20 nm. c, Visualization of S-layer lattices generated by placing back hexamer subunit maps simulated from PDB: 6P5T. Coloring is based on surface curvatures at the center of each subunit. TomoNet overcame the above challenges by utilizing the hexagonal configuration of S-layer lattices. With a minimal manual input, “Auto Expansion” picked over a thousand hexamer S-layer subunits. The intermediate STA result clearly reveals the hexagonal distribution of S-layer inner domains ( Figure 8b ). Visualization of S-layer lattices also shows that the picked particles were arranged in the expected hexagonal pattern, confirming the reliability and applicability of TomoNet as a particle picking tool ( Figure 8c ) and its broad application to structure determination of prokaryotic and archaeal cell walls 45 , 51 . Application to in vitro assembled arrays: nuclear egress complex (NEC) lattice We further validated TomoNet as an integrated high-resolution STA pipeline and an efficient particle picking tool by processing samples containing NEC lattices within budded vehicles. Nuclear egress is a pivotal step in herpes virus replication, driven by NEC and responsible for translocating nascent viral particles from nucleus to cytoplasm. In our reported dataset 52 , NEC heterodimers budded into large vesicles with diameters ranging from 100 nm to 500 nm, forming beehive-like lattices on the inner surface of these vesicles ( Figure 9a, b ). Because of their large sizes, noticeable compressions were observed during the sample freezing, reshaping the vesicles and NEC lattices from spherical to flattened disk shapes ( Figure 9a, b ). This conformational change was a consequence of the limitation in ice thickness imposed by cryoET, which restricts the sample thickness to approximately 250 nm, consequently posing challenges for particle picking. Download figure Open in new tab Figure 9, TomoNet application to in vitro assembled NEC-bound membrane. a, b, Tomographic slice views show a large NEC lattice; the insets show different views of NEC hexamer subunits. Scale bar is 20 nm. c, Orthogonal slice views of an averaged density map generated in TomoNet show that NEC hexamer subunits consist of UL31/UL34 heterodimers. Scale bar is 10 nm. d, Visualization of an NEC lattice generated by placing back averaged maps shows that the large vesicle is compressed into a disk-like shape. The compression caused by sample freezing stretched the lattice, making it flat and split at the air-water surface. Coloring is based on surface curvatures at the center of each subunit. e, Atomic model of the UL31/UL34 heterodimers fits into the final averaged map, with all helices well resolved. TomoNet successfully picked NEC hexamer subunits following the topology of lattices. The intermediate STA result generated in TomoNet already showed the six heterodimers within one hexamer subunit ( Figure 9c ). With these picked particles, high-resolution 3D classifications and refinements were carried out to obtain a final reconstruction of NEC hexamer subunit at 5.4 Å resolution, without preferred orientation bias ( Figure 9c, d and Figure 6c ), and all the helices were well resolved ( Figure 9e ). Visualization of subtomograms placing back shows that the large vesicle was compressed during sample freezing which stretched the NEC lattice, making it appears flat and split at the air-water interface, while the middle part of the lattice appears to be more curved. Application to other types of arrays and free-floating particles The above examples show how TomoNet’s ability to locate particles arrays arranged on flexible spheres (HIV), cell surfaces (S-layer) and nuclear membranes (NEC), which can be considered as topologically 2D lattices. In our published work of various cryoET structures, TomoNet has also been used to locate subtomograms arranged on flexible filaments ( i.e. , 1D arrays) such as the flagella of Trypanosoma brucei 22 , 42 and the amyloid-like sheath protein on β-hoops of the prototypical archaeon, Methanospirillum hungatei 53 . In the case of 3D lattices, TomoNet has been also used to obtain the paraflagellar rod structure of T. brucei 21 . Since TomoNet has integrated packages and is designed for the entire cryoET and STA data processing pipeline, it can also be used as a general-purpose package for subtomogram averaging towards high resolution when particles are free floating and without local order. In the latter case, TomoNet would have the same limitation recognized for all other cryoET software packages, that is, high resolution is currently only achieved for large complexes, such as ribosomes. DISCUSSION In this paper, we report the implementation and application of TomoNet and demonstrate its efficacy in particle picking across three distinct datasets featuring particles with varying lattice configurations. TomoNet stands out as the first software to exhaustively trace lattices following its inherent topology. This unique approach ensures that the particle picking results faithfully reflect in situ or in vitro lattice shape, providing valuable insights into how these lattices are formed by their constituent subunits. For HIV VLPs, TomoNet application enabled us to directly visualize the VLPs lattices and their defects potentially caused by the absence of pentamer subunits. Similarly, for the NEC dataset, TomoNet facilitated a more direct observation of lattice conformation changes resulting from the sample freezing process. Since vesicles in this dataset were too large to be compressed from a sphere into a disk-like shape, the lattice regions near the air-water interface became stretched and subsequently divided into smaller fragments. Moreover, TomoNet demonstrated its exceptional performance, even when dealing with datasets characterized by extremely low contrast. For instance, in the cellular S-layer tomogram of a lamella, S-layer subunits were nearly imperceptible to human observations. Therefore, “Auto Expansion“ excelled in particle picking without requiring denoising or contrast-enhancement algorithms. Additionally, “AI AutoPicking“, the deep learning-based module, demonstrated excellent performance on automatic particle picking, showing potential in handling a wide range of particle types even beyond those with lattice-like arrangements. Compared to the template matching-based “Auto Expansion”, “AI AutoPicking” has several advantages in particle picking. Firstly, it applies to particles situated on flexible lattices and those arranged in scattered patterns, such as cellular ribosomes. The neural network learns to pick by discerning 3D features of individual particles, and it does not require prior knowledge about lattice configuration. Secondly, it utilizes GPUs for fast convolution operations, enabling particle prediction in just several minutes for each tomogram. Thirdly, it does not require the “seed” particles used in “Auto Expansion”, which further reduces human efforts by approximately 5-15 minutes per tomogram. This is especially beneficial for processing extensive tomography datasets with hundreds of tomograms. However, comparing their final output particles, “AI AutoPicking” typically picks fewer particles than “Auto Expansion” because it misses certain particles on the flexible lattices. Thus, these two modules are complementary to each other and can be incorporated to further explore these missing particles. Regarding the pipeline design, each module within TomoNet is designed to be highly independent, ensuring flexibility for integrating future methods and third-party packages. This adaptable framework positions TomoNet as a platform of choice for other developers to build their own innovations. At present, TomoNet is primarily tailored for integration with the Relion-related pipeline. However, it can accommodate specific demands and can be extended to integrate other pipelines, including emClarity 17 , EMAN2 4 , M 54 , and others in the future. In summary, TomoNet significantly simplifies the overall process for users in managing and monitoring every step of the complete cryoET and STA pipeline. Its user-friendly GUI design notably reduces the entry barrier for newcomers to the fast-emerging cryoET field. The particle picking modules of TomoNet provide a general solution for particles organized in lattice-like arrangements, ensuring both accuracy and efficiency, thereby facilitating the high-resolution STA pipeline. METHODS TomoNet is an open-source software package developed using Python. It follows a highly modularized architecture with each module responsible for specific tasks in a typical cryoET and STA data processing pipeline. Modules in TomoNet mainly cover the upper stream of the cryoET and STA pipeline including procedures of motion correction, tilt series generation, tomogram reconstruction, CTF estimation and particle picking, while leave the high-resolution 3D refinement to established software package like Relion ( Figure 1 ). The design of a modern GUI, established with PyQt5 platform, enhances user-friendliness, and helps with tracking the processing progress ( Figure 2 ). With table views, users can obtain a comprehensive overview of the entire dataset, facilitating direct and intuitive management for each tomogram ( Figure 2 ). Implementation of modules for motion correction, tomogram reconstruction and CTF estimation Motion correction, tomogram reconstruction, and CTF estimation related functions are organized into individual modules in TomoNet, with the integration of corresponding external software packages including MotionCorr2 36 , IMOD 37 or AreTomo 38 and CTFFIND4 39 , respectively. Since their codes are not rewritten in TomoNet, users have to install each of them before using the corresponding modules. The “Motion Correction” module is used to correct bean-induced sample motion. It requires an input folder path that contains all the dose fractionated frames, then user can specify their MotionCorr2 parameters in the GUI. After clicking the “RUN” button, TomoNet will perform motion correction for all the input images and save the results in a separated directory. This module also allows on-the-fly motion correction during data collection. The “3D Reconstruction” module comprises two sub-functions: “TS Generation” and “Reconstruction”. Within “TS Generation”, users can readily assemble tilt series for each tomogram from the previously generated motion corrected images. It provides advanced options for data cleaning, such as setting a minimum acceptable number of tilt images for a tomogram, removing duplicate images at the same tilt angle by excluding images with older time stamps. The “Reconstruction” tab automatically reads and lists all tomograms in a table view, with essential information, such as tilt image number and alignment errors, and action buttons for restart, continue and delete individual tomogram reconstruction process. This simplifies the assessment of reconstruction results and facilitating tomogram reconstruction management. The “CTF Estimation“ module is used for the tilt series defocus estimation, with support of parallel processing using multiple CPUs. Its outcomes are also listed in a table view with visualization features, such as displaying defocus at 0 degree and plotting the defocus distribution across all tilt angles. Implementation of the “Manual Picking” module The “Manual Picking” module is designed for general management of manual particle picking, especially for the preparation of “seed” particles required in “Auto Expansion”. IMOD stalkInit picking criteria is implemented to define the Y-axis for each particle with 2 points, and the center in between them. In the example of HIV dataset, 5-10 particles were manually picked as the “seed” particles for each VLP lattice, which only takes several minutes per tomogram ( Figure 5a ). Design and implementation of the “Auto Expansion” module “Auto Expansion” consists of three steps as shown in Figure 2 . “Generate tomograms.star” is used to generate a STAR format file that maintain information of tomograms and their associated “seed” particles to be applied in “Auto Expansion”. “Generate Picking Parameter” is used to set up parameters required for particle set expansion through the described iterative process. The parameters include angular search ranges and steps, translational search ranges and steps, a “transition list” (explained later), box size used in particle alignment, distance between neighboring repeating subunits, reference and mask map, cross-correlation threshold, etc. The “transition list” is customized by users to describe the targeting lattice configuration, with each transition denoted by [sx, sy, sz] , where sx, sy and sz are translational shifts from the center of “seed” particle to one of its neighbors along X, Y and Z-axis, respectively. Thus, “Auto Expansion” can use it to guide the search of “candidate” particles. These user defined parameters will then be saved into a JSON format file. “Run Particle Expansion” takes the above STAR and JSON format files as inputs to perform the iterative particle set expansion. During the “Auto Expansion” processing, three directories will be generated for each tomogram. They are “ TomoName ” as the working directory for carrying out the current iteration, “ TomoName_cache ” that stores intermediate results from finished iterations, and “ TomoName_final ” that stores the final particle picking results. The iteration number of “Auto Expansion” is typically greater than one. However, “Auto Expansion” allows for some special usage cases. For example, in the scenario when users need to modify the particle picking setting such as a different cross-correlation threshold, user can generate the new picking parameter file, then execute “Run Particle Expansion” by setting the iteration number as 0. This prompts the program to skip the “candidate” searching steps, but just gather all intermediate results saved in “ TomoName_cache ” directories, then generate a new “ TomoName_final ” result. Design and implementation of the “AI AutoPicking” module The “AI AutoPicking” module comprise 3 main steps, “Prepare Training Dataset”, “Train Neural Network” and “Predict Particles coordinates”. It uses supervised machine learning that requires users to provide ground truth, i.e. , tomogram with the associated particle coordinates files, for the model training. In this study, the ground truth data were prepared by “Auto Expansion”. In “Prepare Training Dataset”, extracted subtomograms are used as inputs to the network training model for two reasons. Firstly, the size of tomogram used for picking is typically around 1000×1000×1000 voxels which is not applicable to be loaded in the GPU memory, but the size of extracted subtomograms is under 100×100×100 voxels. Secondly, it helps with increasing the number of training data pairs to avoid over-fitting during the network training. For the model output, the particle coordinates information was embedded into 3D binary segmentation maps, where the voxels associated with particles were set to 1, otherwise set to 0 ( Figure 4a ). In “Train Neural Network”, the above extracted subtomograms paired with their associated segmentation maps are used to train a neural network model to be a binary classifier that predict whether a voxel is near the center of a particle. The network architecture employed is derived from the one used in IsoNet 46 as it is well-suited for capturing generalized features of 3D objects ( Figure 4b ). Since the learning task is voxel-wisely binary classification, cross entropy loss function is used instead of minimum squared error (MSE). Equipped with one RTX 3080Ti graphic card, the training process can be completed swiftly within 1-2 hours if using the default parameters. In “Predict Particles coordinates”, users can apply the trained model on the entire tomography dataset for particle coordinate prediction ( Figure 4c ). For each tomogram, TomoNet generate a predicted segmentation map first, then its particle coordinates information can be retrieved from the segmentation map by utilizing the hierarchical clustering algorithm from scipy module in Python. Implementation of tools within the “Other Utilities” module The “Other Utilities” module consists of two sub-functions: “Recenter | Rotate | Assemble to .star file“ and “3D Subtomogram Place Back“ as useful tools for post particle picking processing. The first one allows users to assemble and convert the particle picking results into a STAR format file following the Relion4 convention, reset particles center to its symmetric center, and align the rotation axis to Relion Z-axis. The second one takes a user-provided STAR format file that contains particles information as input, then generates a ChimeraX 55 session file for 3D subtomograms placing back and a clean version of STAR format file with “bad” particles removed. This not only allows users to validate the accuracy of particle picking before importing into Relion, but also enables direct observation of the distribution and configuration of subunits after the high-resolution 3D refinements, providing overall in situ lattice observations ( Figure 7 ). Processing tomograms of HIV VLP dataset The HIV VLP dataset was downloaded from the Electron Microscopy Public Image Archive (EMPIAR) with the accession code EMPIAR-10164 43 . Four tilt series, TS_01, TS_43, TS_45 and TS_54, were used in this study. Downloaded micrographs were loaded into the TomoNet pipeline to perform tilt series assembly, CTF estimation, and tomogram reconstruction using the WBP algorithm. Four-time binned tomograms with 5.4 Å pixel size were used for further particle picking. Firstly, tomograms TS_01 and TS_43 were used for “seed“ particles preparation on 3 selected VLPs per tomogram, and an initial reference map was generated by averaging them in PEET. Secondly, one run of “Auto Expansion” was applied on the above two tomograms to get more particles, such as to refine the reference. Thirdly, with an improved reference, a new run of “Auto Expansion” was applied on the selected 3 VLPs in both tomogram ( Figure 5b ), then the particle picking result was used for neural network training in “AI AutoPicking”. Fourthly, after the particle prediction on all four tomograms with a trained model, “AI AutoPicking” produced 4,860, 3,704, 4,550 and 2,101 particles for tomograms TS_01, TS_43, TS_45 and TS_54, as shown in Figure 5c . Lastly, the predicted particles were input as “seed” particles for the final run of “Auto Expansion”, resulting in 5,765, 4,043, 5,006, and 2,838 particles for tomograms TS_01, TS_43, TS_45 and TS_54, which were imported into Relion to perform high-resolution refinements. Following the same procedure carried out in the Relion4 tutorial together with TomoNet 3D classification, the Gag hexamer structure was resolved at 3.2 Å resolution with 13,558 particles from four tomograms. Resolution was calculated in Relion and on 3DFSC Processing Server 56 . The global resolution reported is based on the “gold standard” refinement procedures and the 0.143 Fourier shell correlation (FSC) criterion ( Figure 6 ). Processing one tomogram of C. Crescentus S-layer The FIB-milled C. crescentus data of one reconstructed tomogram was downloaded from Electron Microscopy Data Bank (EMDB) with the accession code EMD-23622 50 . This tomogram was directly used for “seed” particles preparation on two of the cells. Around 30 “seed” particles were manually picked and averaged using PEET to generate an initial reference map. “Auto Expansion” was applied on the “seed” particles for 5 iterations to get more particles such as to refine the reference map. With the improved reference map, another run of “Auto Expansion” was applied to the same “seed” particles for 15 iterations to search all particles on the outer surface of the cells, and finally yielded ∼1,500 S-layer particles of hexamer subunits ( Figure 8c ). Processing tomograms of NEC budding in vitro The cryoET grid preparation and data collection were previously described 52 . Motion correction, tomogram reconstruction and CTF estimation were performed using TomoNet. Around 50-150 “seed” particles were manually picked for each tomogram. “Auto Expansion” were applied on a total of 35 tomograms and yield the ∼48,000 particles before Relion refinements. Following one round of 3D auto-refine job under four-binned pixel size and several rounds of 3D auto-refine jobs under two-binned pixel size and one round of 3D auto-refine under unbinned pixel size, together with TomoNet 3D classifications, the NEC hexamer structure was resolved at 5.4 Å resolution with totally 35,039 particles. 3D visualization IMOD 37 was used to visualize the 2D tomographic and segmentation map slices. UCSF ChimeraX 55 was used to visualize the STA results and the lattices generated by 3D subtomogram place back. The atomic models were fitted into the density map using the “fit in map” tool in ChimeraX. AVAILABILITY TomoNet code is available on Github website at https://github.com/logicvay2010/TomoNet , with a user manual. For the HIV VLPs dataset, the raw data was downloaded from the Electron Microscopy Public Image Archive (EMPIAR) with accession code EMPIAR-10164 43 , the Gag atomic model was downloaded from the Protein Data Bank (PDB) with accession code 5L93 43 . For the C. Crescentus S-layer dataset, the reconstructed tomogram was downloaded from the Electron Microscopy Data Bank (EMDB) with accession code EMD-23622 50 , and the subunit model was generated using atomic model with PDB accession code 6P5T 57 . The STA result of NEC hexamer is from EMDB with accession code EMD-40224 52 . The other data that support the findings of this study are available from the corresponding authors upon reasonable request. AUTHORSHIP CONTRIBUTIONS HW and ZHZ initialized and ZHZ supervised research; HW wrote the code and developed the software GUI with help from SL; HW, SL and XY tested the software on different datasets; HW, and ZHZ wrote the manuscript; JZ and XY assisted the manuscript writing; all authors reviewed and approved the paper. COMPETING INTERESTS STATEMENT The authors declare that there is no conflict of interest. Movie Legends Movie 1, A spherical VLP consisting of hexamer Gag subunits, colored by local surface curvature. “bad” particles with wrong alignment are shown as red. Movie 2, A VLP lattice with irregular shape. ACKNOWLEDGEMENTS We thank Elizabeth Draganova and Ekaterina Heldwein for the NEC dataset. This project is supported by grants from the US National Institutes of Health (GM071940 to Z.H.Z.) and the National Science Foundation (DMR-1548924 to Z.H.Z.). Footnotes https://github.com/logicvay2010/TomoNet REFERENCES ↵ Kimanius , D. , Dong , L. , Sharov , G. , Nakane , T. & Scheres , S. H. W . New tools for automated cryo-EM single-particle analysis in RELION-4.0 . Biochem J 478 , 4169 – 4185 , doi: 10.1042/BCJ20210708 ( 2021 ). OpenUrl CrossRef ↵ Punjani , A. , Rubinstein , J. L. , Fleet , D. J. & Brubaker , M. A . cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination . Nat Methods 14 , 290 – 296 , doi: 10.1038/nmeth.4169 ( 2017 ). OpenUrl CrossRef PubMed ↵ Wan , W. & Briggs , J. A. Cryo-Electron Tomography and Subtomogram Averaging . Methods Enzymol 579 , 329 – 367 , doi: 10.1016/bs.mie.2016.04.014 ( 2016 ). OpenUrl CrossRef ↵ Chen , M. et al. A complete data processing workflow for cryo-ET and subtomogram averaging . Nat Methods 16 , 1161 – 1168 , doi: 10.1038/s41592-019-0591-8 ( 2019 ). OpenUrl CrossRef Zhang , P . Advances in cryo-electron tomography and subtomogram averaging and classification . Curr Opin Struct Biol 58 , 249 – 258 , doi: 10.1016/j.sbi.2019.05.021 ( 2019 ). OpenUrl CrossRef Castano-Diez , D. & Zanetti , G . In situ structure determination by subtomogram averaging . Curr Opin Struct Biol 58 , 68 – 75 , doi: 10.1016/j.sbi.2019.05.011 ( 2019 ). OpenUrl CrossRef Hong , Y. , Song , Y. , Zhang , Z. & Li , S . Cryo-Electron Tomography: The Resolution Revolution and a Surge of In Situ Virological Discoveries . Annu Rev Biophys 52 , 339 – 360 , doi: 10.1146/annurev-biophys-092022-100958 ( 2023 ). OpenUrl CrossRef Huang , Y. , Zhang , Y. & Ni , T . Towards in situ high-resolution imaging of viruses and macromolecular complexes using cryo-electron tomography . J Struct Biol 215 , 108000 , doi: 10.1016/j.jsb.2023.108000 ( 2023 ). OpenUrl CrossRef Sibert , B. S. et al. Workflow for High-resolution Sub-volume Averaging from Heterogenous Viral and Virus-like Assemblies . Microsc Microanal 29 , 943 – 944 , doi: 10.1093/micmic/ozad067.470 ( 2023 ). OpenUrl CrossRef ↵ Kopylov , M. , Bobe , D. , Johnston , J. D. & Paraan , R. M . Modern Tools for In-situ Tomography . Microsc Microanal 29 , 954 – 955 , doi: 10.1093/micmic/ozad067.476 ( 2023 ). OpenUrl CrossRef ↵ Ni , T. et al. High-resolution in situ structure determination by cryo-electron tomography and subtomogram averaging using emClarity . Nat Protoc 17 , 421 – 444 , doi: 10.1038/s41596-021-00648-5 ( 2022 ). OpenUrl CrossRef Xue , L. et al. Visualizing translation dynamics at atomic detail inside a bacterial cell . Nature 610 , 205 – 211 , doi: 10.1038/s41586-022-05255-2 ( 2022 ). OpenUrl CrossRef ↵ Zivanov , J. et al. A Bayesian approach to single-particle electron cryo-tomography in RELION-4.0 . Elife 11 , doi: 10.7554/eLife.83724 ( 2022 ). OpenUrl CrossRef ↵ Obr , M. & Schur , F. K. M. in Advances in Virus Research Vol. 105 (ed Félix A. Rey ) 117 – 159 ( Academic Press , 2019 ). OpenUrl CrossRef ↵ Castano-Diez , D. , Kudryashev , M. , Arheit , M. & Stahlberg , H . Dynamo: a flexible, user-friendly development tool for subtomogram averaging of cryo-EM data in high-performance computing environments . J Struct Biol 178 , 139 – 151 , doi: 10.1016/j.jsb.2011.12.017 ( 2012 ). OpenUrl CrossRef PubMed ↵ Scaramuzza , S. & Castaño-Díez , D . Step-by-step guide to efficient subtomogram averaging of virus-like particles with Dynamo . PLOS Biology 19 , e3001318 , doi: 10.1371/journal.pbio.3001318 ( 2021 ). OpenUrl CrossRef ↵ Himes , B. A. & Zhang , P . emClarity: software for high-resolution cryo-electron tomography and subtomogram averaging . Nature Methods 15 , 955 – 961 , doi: 10.1038/s41592-018-0167-z ( 2018 ). OpenUrl CrossRef PubMed ↵ Wagner , T. et al. SPHIRE-crYOLO is a fast and accurate fully automated particle picker for cryo-EM . Commun Biol 2 , 218 , doi: 10.1038/s42003-019-0437-z ( 2019 ). OpenUrl CrossRef PubMed ↵ Bharat , T. A. M. & Scheres , S. H. W . Resolving macromolecular structures from electron cryo-tomography data using subtomogram averaging in RELION . Nature Protocols 11 , 2054 – 2065 , doi: 10.1038/nprot.2016.124 ( 2016 ). OpenUrl CrossRef PubMed ↵ Tegunov , D. & Cramer , P . Real-time cryo-electron microscopy data preprocessing with Warp . Nature Methods 16 , 1146 – 1152 , doi: 10.1038/s41592-019-0580-y ( 2019 ). OpenUrl CrossRef PubMed ↵ Zhang , J. et al. Structure of the trypanosome paraflagellar rod and insights into non-planar motility of eukaryotic cells . Cell Discov 7 , 51 , doi: 10.1038/s41421-021-00281-2 ( 2021 ). OpenUrl CrossRef ↵ Imhof , S. et al. Cryo electron tomography with volta phase plate reveals novel structural foundations of the 96-nm axonemal repeat in the pathogen Trypanosoma brucei . eLife 8 , e52058 , doi: 10.7554/eLife.52058 ( 2019 ). OpenUrl CrossRef ↵ Si , Z. et al. Different functional states of fusion protein gB revealed on human cytomegalovirus by cryo electron tomography with Volta phase plate . PLOS Pathogens 14 , e1007452 , doi: 10.1371/journal.ppat.1007452 ( 2018 ). OpenUrl CrossRef PubMed ↵ Bharat , T. A. et al. Cryo-electron tomography of Marburg virus particles and their morphogenesis within infected cells . PLoS Biol 9 , e1001196 , doi: 10.1371/journal.pbio.1001196 ( 2011 ). OpenUrl CrossRef PubMed ↵ Grünewald , K. et al. Three-dimensional structure of herpes simplex virus from cryo-electron tomography . Science 302 , 1396 – 1398 , doi: 10.1126/science.1090284 ( 2003 ). OpenUrl Abstract / FREE Full Text ↵ Zanetti , G. et al. The structure of the COPII transport-vesicle coat assembled on membranes . Elife 2 , e00951 , doi: 10.7554/eLife.00951 ( 2013 ). OpenUrl CrossRef PubMed ↵ Böhm , J. et al. Toward detecting and identifying macromolecules in a cellular context: Template matching applied to electron tomograms . Proceedings of the National Academy of Sciences 97 , 14245 – 14250 , doi : doi: 10.1073/pnas.230282097 ( 2000 ). OpenUrl Abstract / FREE Full Text ↵ de Teresa-Trueba , I. et al. Convolutional networks for supervised mining of molecular patterns within cellular context . Nature Methods 20 , 284 – 294 , doi: 10.1038/s41592-022-01746-2 ( 2023 ). OpenUrl CrossRef Moebel , E. et al. Deep learning improves macromolecule identification in 3D cellular cryo-electron tomograms . Nature Methods 18 , 1386 – 1394 , doi: 10.1038/s41592-021-01275-4 ( 2021 ). OpenUrl CrossRef Wu , S. , Liu , G. & Yang , G. in 2022 IEEE 19th International Symposium on Biomedical Imaging (ISBI) . 1 – 5 . ↵ Hao , Y. et al. VP-Detector: A 3D multi-scale dense convolutional neural network for macromolecule localization and classification in cryo-electron tomograms . Computer Methods and Programs in Biomedicine 221 , 106871 , doi: 10.1016/j.cmpb.2022.106871 ( 2022 ). OpenUrl CrossRef ↵ Rice , G. et al. TomoTwin: generalized 3D localization of macromolecules in cryo-electron tomograms with structural data mining . Nature Methods 20 , 871 – 880 , doi: 10.1038/s41592-023-01878-z ( 2023 ). OpenUrl CrossRef ↵ Balyschew , N. et al. Streamlined structure determination by cryo-electron tomography and subtomogram averaging using TomoBEAR . Nature Communications 14 , 6543 , doi: 10.1038/s41467-023-42085-w ( 2023 ). OpenUrl CrossRef ↵ Jimenez de la Morena , J. , et al. ScipionTomo: Towards cryo-electron tomography software integration, reproducibility, and validation . J Struct Biol 214 , 107872 , doi: 10.1016/j.jsb.2022.107872 ( 2022 ). OpenUrl CrossRef ↵ Liu , H.-F. et al. nextPYP: a comprehensive and scalable platform for characterizing protein variability in situ using single-particle cryo-electron tomography . Nature Methods , doi: 10.1038/s41592-023-02045-0 ( 2023 ). OpenUrl CrossRef ↵ Zheng , S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy . Nat Methods 14 , 331 – 332 , doi: 10.1038/nmeth.4193 ( 2017 ). OpenUrl CrossRef PubMed ↵ Kremer , J. R. , Mastronarde , D. N. & McIntosh , J. R . Computer visualization of three-dimensional image data using IMOD . J Struct Biol 116 , 71 – 76 , doi: 10.1006/jsbi.1996.0013 ( 1996 ). OpenUrl CrossRef PubMed Web of Science ↵ Zheng , S. et al. AreTomo: An integrated software package for automated marker-free, motion-corrected cryo-electron tomographic alignment and reconstruction . Journal of Structural Biology: X 6 , 100068 , doi: 10.1016/j.yjsbx.2022.100068 ( 2022 ). OpenUrl CrossRef ↵ Rohou , A. & Grigorieff , N . CTFFIND4: Fast and accurate defocus estimation from electron micrographs . J Struct Biol 192 , 216 – 221 , doi: 10.1016/j.jsb.2015.08.008 ( 2015 ). OpenUrl CrossRef PubMed ↵ Heumann , J. M. , Hoenger , A. & Mastronarde , D. N . Clustering and variance maps for cryo-electron tomography using wedge-masked differences . Journal of Structural Biology 175 , 288 – 299 , doi: 10.1016/j.jsb.2011.05.011 ( 2011 ). OpenUrl CrossRef PubMed ↵ Hall , S. R . The STAR file: a new format for electronic data transfer and archiving . Journal of Chemical Information and Computer Sciences 31 , 326 – 333 , doi: 10.1021/ci00002a020 ( 1991 ). OpenUrl CrossRef Web of Science ↵ Shimogawa , M. M. et al. FAP106 is an interaction hub for assembling microtubule inner proteins at the cilium inner junction . Nature Communications 14 , 5225 , doi: 10.1038/s41467-023-40230-z ( 2023 ). OpenUrl CrossRef ↵ Schur , F. K. M. et al. An atomic model of HIV-1 capsid-SP1 reveals structures regulating assembly and maturation . Science 353 , 506 – 508 , doi : doi: 10.1126/science.aaf9620 ( 2016 ). OpenUrl Abstract / FREE Full Text ↵ von Kügelgen , A. , Alva , V. & Bharat , T. A. M . Complete atomic structure of a native archaeal cell surface . Cell Reports 37 , 110052 , doi: 10.1016/j.celrep.2021.110052 ( 2021 ). OpenUrl CrossRef ↵ Pum , D. , Breitwieser , A. & Sleytr , U. B . Patterns in Nature—S-Layer Lattices of Bacterial and Archaeal Cells . Crystals 11 , 869 ( 2021 ). ↵ Liu , Y.-T. et al. Isotropic reconstruction for electron tomography with deep learning . Nature Communications 13 , 6482 , doi: 10.1038/s41467-022-33957-8 ( 2022 ). OpenUrl CrossRef PubMed ↵ Tan , A. , Pak , A. J. , Morado , D. R. , Voth , G. A. & Briggs , J. A. G . Immature HIV-1 assembles from Gag dimers leaving partial hexamers at lattice edges as potential substrates for proteolytic maturation . Proceedings of the National Academy of Sciences 118 , e2020054118 , doi : doi: 10.1073/pnas.2020054118 ( 2021 ). OpenUrl Abstract / FREE Full Text ↵ Guo , S. , Saha , I. , Saffarian , S. & Johnson , M. E . Structure of the HIV immature lattice allows for essential lattice remodeling within budded virions . eLife 12 , e84881 , doi: 10.7554/eLife.84881 ( 2023 ). OpenUrl CrossRef ↵ Talledge , N. et al. HIV-2 Immature Particle Morphology Provides Insights into Gag Lattice Stability and Virus Maturation . Journal of Molecular Biology 435 , 168143 , doi: 10.1016/j.jmb.2023.168143 ( 2023 ). OpenUrl CrossRef ↵ Lasker , K. et al. The material properties of a bacterial-derived biomolecular condensate tune biological function in natural and synthetic systems . Nature Communications 13 , 5643 , doi: 10.1038/s41467-022-33221-z ( 2022 ). OpenUrl CrossRef ↵ Sleytr , U. B. , Schuster , B. , Egelseer , E. M. & Pum , D. S-layers: principles and applications . FEMS Microbiol Rev 38 , 823 – 864 , doi: 10.1111/1574-6976.12063 ( 2014 ). OpenUrl CrossRef PubMed ↵ Draganova , E. B. et al. The universal suppressor mutation in the HSV-1 nuclear egress complex restores membrane budding defects by stabilizing the oligomeric lattice . bioRxiv , 2023.2006.2022.546118 , doi: 10.1101/2023.06.22.546118 ( 2023 ). OpenUrl Abstract / FREE Full Text ↵ Wang , H. et al. Hierarchical organization and assembly of the archaeal cell sheath from an amyloid-like protein . Nature Communications 14 , 6720 , doi: 10.1038/s41467-023-42368-2 ( 2023 ). OpenUrl CrossRef ↵ Tegunov , D. , Xue , L. , Dienemann , C. , Cramer , P. & Mahamid , J . Multi-particle cryo-EM refinement with M visualizes ribosome-antibiotic complex at 3.5 Å in cells . Nature Methods 18 , 186 – 193 , doi: 10.1038/s41592-020-01054-7 ( 2021 ). OpenUrl CrossRef ↵ Meng , E. C. et al. UCSF ChimeraX: Tools for Structure Building and Analysis . Protein Sci , e4792 , doi: 10.1002/pro.4792 ( 2023 ). OpenUrl CrossRef PubMed ↵ Tan , Y. Z. et al. Addressing preferred specimen orientation in single-particle cryo-EM through tilting . Nat Methods 14 , 793 – 796 , doi: 10.1038/nmeth.4347 ( 2017 ). OpenUrl CrossRef PubMed ↵ Herrmann , J. et al. A bacterial surface layer protein exploits multistep crystallization for rapid self-assembly . Proceedings of the National Academy of Sciences 117 , 388 – 394 , doi : doi: 10.1073/pnas.1909798116 ( 2020 ). OpenUrl Abstract / FREE Full Text View the discussion thread. Back to top Previous Next Posted February 19, 2024. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following TomoNet: A streamlined cryoET software pipeline with automatic particle picking on flexible lattices Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share TomoNet: A streamlined cryoET software pipeline with automatic particle picking on flexible lattices Hui Wang , Shiqing Liao , Xinye Yu , Jiayan Zhang , Z. Hong Zhou bioRxiv 2024.02.17.580557; doi: https://doi.org/10.1101/2024.02.17.580557 Share This Article: Copy Citation Tools TomoNet: A streamlined cryoET software pipeline with automatic particle picking on flexible lattices Hui Wang , Shiqing Liao , Xinye Yu , Jiayan Zhang , Z. Hong Zhou bioRxiv 2024.02.17.580557; doi: https://doi.org/10.1101/2024.02.17.580557 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Cell Biology Subject Areas All Articles Animal Behavior and Cognition (7652) Biochemistry (17752) Bioengineering (13937) Bioinformatics (42088) Biophysics (21502) Cancer Biology (18658) Cell Biology (25587) Clinical Trials (138) Developmental Biology (13410) Ecology (19950) Epidemiology (2067) Evolutionary Biology (24381) Genetics (15640) Genomics (22567) Immunology (17780) Microbiology (40505) Molecular Biology (17219) Neuroscience (88830) Paleontology (667) Pathology (2846) Pharmacology and Toxicology (4840) Physiology (7667) Plant Biology (15182) Scientific Communication and Education (2048) Synthetic Biology (4307) Systems Biology (9840) Zoology (2274)