Cryo-electron tomography (cryo-ET) is an imaging technique rapidly gaining popularity for the direct visualisation of unique biological objects in 3D. Repeating structural motifs present in cryo-ET data can be reconstructed at higher resolution by subtomogram averaging (STA), allowing for the possibility of studying macromolecular structure in situ (Schur, 2019) . STA has developed alongside single particle analysis (SPA) in cryo-electron microscopy (cryo-EM), a technique which has benefited significantly in the last ten years from advances in both electron detection hardware and image processing software (Bai et al., 2015; Nakane et al., 2020).

State-of-the-art subtomogram averaging workflows often co-opt tools and adapt ideas from single particle analysis for tomography and subtomogram averaging (Sanchez et al., 2020; Turoňová et al., 2020) . An unfortunate consequence of this side-by-side development is a somewhat fragmented software ecosystem which employs various file formats and metadata conventions (Leigh et al., 2019). Despite the advent of complete or near complete integrated solutions for cryo-ET and STA (Chen et al., 2019; Himes and Zhang, 2018; Tegunov et al., 2021) , employing optimal methodology for a given dataset often requires the creative combination of different approaches which may not all be present within a single integrated pipeline. For those new to cryo-ET, the burden of interfacing many different software packages in a complex workflow often represents a barrier to the testing of alternative approaches.

Image processing for STA from cryo-ET data is a complex process which can broadly be divided into three blocks (Figure 1). The first, ‘preprocessing’, transforms experimental 2D micrographs into 3D reconstructions of an imaged region. The second, ’particle picking’, generates putative positions and orientations for objects of interest within each volume as well as initial reference(s) for subsequent refinement. Finally, the ‘refinement’ block is concerned with the optimisation of reconstruction(s) from imaging data associated with each particle position. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The preprocessing block encompasses all steps in the generation of tomograms from experimental data. Experimental cryo-ET data is usually acquired as a set of multi-frame 2D micrographs, one per tilt angle in a tilt-series. Typically, this block includes per-tilt inter-frame motion estimation and correction, contrast transfer function (CTF) estimation, tilt-series alignment and 3D reconstruction (Tegunov and Cramer, 2019) . Tilt-series alignment is the least automated of these steps, with significant time often spent optimising alignments in an attempt to produce more accurate reconstructions.

Particle picking can be somewhat entangled with the process of initial model generation, with some particle picking methods requiring an initial model and others being reference-free. Exhaustive, reference-based template matching approaches are widely accepted as an imperfect solution for particle picking due to the computational overhead and the need for subsequent dataset cleaning (Basanta et al., 2020; Schaffer et al., 2019) . Care must also be taken to avoid model bias. Particle 60 picking methods based on deep-learning have been proposed, which may help to address some of 61 these limitations (Chen et al., 2017; Moebel et al., 2021). Another class of particle picking methods 62 derive putative particle positions and orientations from a 3D model of a supporting geometry, usually 63 generated from minimal manual annotations (Castaño-Díez et al., 2017). These methods can be used 64 to impose prior knowledge about how particle poses relate to the supporting geometry during 65 refinement. Employing such a priori knowledge is advantageous, reducing both the computational 66 burden of global searches and the likelihood of ending up in incorrect local minima during refinement. 67 These advantages come at the expense of extra time spent on manual annotation and management 68 of more metadata. 69 Refinement has typically meant the optimisation of particle poses within fixed 3D 70 reconstructions (Wan and Briggs, 2016). More recently, this block has been extended to include 71 procedures for the optimisation of many geometrical and electron-optical parameters, improving the 72 resolutions attainable by STA (Bartesaghi et al., 2012; Chen et al., 2019; Himes and Zhang, 2018; 73 Tegunov et al., 2021) . 74 Since the advent of methods for 3D contrast transfer function (CTF) correction (Turoňová et 75 al., 2017) and a posteriori improvement of an existing set of alignment parameters (Bartesaghi et al., 76 2012; Chen et al., 2019; Himes and Zhang, 2018; Tegunov et al., 2021) , the potential for studying 77 macromolecular structure at intermediate resolution (3-5 Å) by STA is becoming a reality. Despite 78 these advances, there remain relatively few examples of STA reconstructions in this resolution regime 79 in the Electron Microscopy Data Bank (EMDB), indicating that this work remains non-trivial (Figure 180 figure supplement 1). 81 Dynamo is a flexible, extensible software environment for subtomogram averaging, providing 82 a variety of powerful tools for data management, annotation and analysis (Castaño-Díez et al., 2012). 83 The Warp-RELION-M pipeline is an integrated solution for cryo-ET and SPA, yielding the highest 84 resolution reconstructions from cryo-ET data thus far (Tegunov et al., 2021) . In the materials and 85 methods section, we describe tools designed to facilitate certain computational aspects of cryo-ET by 86 simplifying metadata handling, increasing automation and interfacing Dynamo with the Warp87 RELION-M pipeline. 88 In the results and discussion section, we show that combination of these tools enables 89 working within a flexible framework for ab initio and geometrical approaches to subtomogram 90 averaging. We illustrate the power of working within such a framework on two datasets. We also 91 provide a collaborative, online platform for the community to share knowledge about cryo-ET data 92 processing, to which we contribute a complete guide for achieving state-of-the-art results on a five 93 tomogram subset of EMPIAR-10164. 94 Materials and Methods 95 Integrating on-the-fly tilt-series alignment and data management into the Warp preprocessing 96 pipeline 97 Tomogram reconstruction within the Warp preprocessing pipeline currently requires the use of 98 an external software package, IMOD (Kremer et al., 1996), for tilt-series alignment. In the absence of 99 a fully integrated solution in the currently available version of Warp (1.0.9), we developed 100 autoalign_dynamo, a package for automating fiducial-based tilt-series alignment using Dynamo and 101 IMOD. dautoalign4warp provides a simple command line interface for on-the-fly alignment of tilt102 series. The program proceeds by dynamic generation of a Dynamo tilt-series alignment workflow with 103 appropriate parameters derived from minimal user input. The workflow is executed, generating a set 104 of refined fiducial positions. Refined fiducial positions are converted using dms2mod (provided by 105 autoalign_dynamo) and used to generate tilt-series alignment parameters using the IMOD program 106 tiltalign. Key tiltalign parameters fixed within this procedure are: solve for one tilt axis for the entire tilt107 series, fix tilt-angles at their nominal values, use robust fitting for parameter estimation. All 108 parameters can be seen and modified in the tiltalign_wrapper script. All data is subsequently 109 organised such that alignments can be imported directly into Warp for tomogram reconstruction, 110 requiring no further user input. The base function autoalign which wraps Dynamo tilt-series alignment 111 functionality is provided for those who wish to apply this procedure outside of the Warp pipeline. A 112 function warp2catalogue serves to set up a database called a Dynamo catalogue for tomograms 113 reconstructed in Warp. The catalogue is set up such that all visualisation operations make use of a 114 deconvolved reconstruction, filtered for optimal visualisation, whilst all particle extraction operations 115 make use of the corresponding unfiltered volume. 116 Dynamo tilt-series alignment workflows were released recently and as such are not yet 117 described in the literature, although details and the source code are available in the public domain 118 (Castaño-Díez, 2019) . To facilitate the readers understanding, we provide a brief overview of the 119 alignment algorithm, noting that it was not implemented by us. A binary, synthetic template of a 120 fiducial marker is generated based on user input and used to detect candidate fiducial positions within 121 a tilt-series by cross-correlation. 300 sub-images are extracted from the tilt-series at the peaks of the 122 resulting cross-correlation matrix and averaged to produce a template for detecting fiducial markers in 123 the data. The template is used to detect initial fiducial positions in the tilt-series by cross-correlation. 124 Cross-correlation peaks are analysed and those observations not meeting a minimum degree of 125 rotational symmetry are discarded. Observations are indexed to link observations of the same fiducial 126 marker in multiple micrographs by pairwise cross-correlation of the observations between 127 neighbouring micrographs. The longest ‘trails’ of linked observations are used to generate an initial 128 3D model of fiducial positions. The 3D model is iteratively reindexed by reprojection of fiducial 129 positions against the tilt-series, adding observations which fall within a distance threshold to the set of 130 observations to be included for further refinement. Tilt-images lacking at this point are reintegrated by 131 a procedure comparing the reprojection of the current model with the cloud of initial observations 132 found on that micrograph. The number of observations is maximised by the reintegration of missing 133 fiducial markers from other images using the same procedure. The projection model is iteratively 134 refined before the positions of fiducial markers are independently iteratively refined against an 135 average of all observations of that marker. The final set of refined markers is pruned according to the 136 root mean square deviation (RMSD) between measured fiducial position and the reprojected position 137 of the 3D model. 138 Python package for Euler angle conversions 139 Euler angles are used by many cryo-EM software packages to describe the orientation of a 140 rigid body with respect to a fixed coordinate system. eulerangles is a self-contained Python package 141 which provides a simple application programming interface (API) for the batch conversion of Euler 142 angles into rotation matrices, rotation matrices into Euler angles and interconversion of Euler angles 143 defined according to different conventions. Conversions can take place between all possible 144 definitions of Euler angles in a right-handed coordinate system and the implementation is vectorised, 145 allowing it to handle large datasets efficiently. Conventions for Euler angles from Dynamo and Warp146 RELION-M are provided in the library itself, along with a simple mechanism for the definition of 147 conventions from other software packages. 148 Python packages for metadata handling 149 We have developed two self-contained Python packages starfile and dynamotable to facilitate 150 handling the primary metadata systems of RELION (Scheres, 2012) and Dynamo (Castaño-Díez et 151 al., 2012). These packages provide input/output functionality via a simple API. Data is exposed as 152 pandas DataFrame objects, effectively interfacing these metadata systems directly with the scientific 153 Python (Virtanen et al., 2020) ecosystem. 154 Integration of data collected in Tomography 5 as multi-frame micrographs into the Warp 155 preprocessing pipeline 156 In the Warp cryo-ET preprocessing pipeline, only metadata from the SerialEM data collection 157 program in the form of mdoc files is currently supported. Thermo Scientific Tomography 5 (Tomo 5) is 158 the official solution provided with Thermo Scientific microscopes for electron tomography experiments 159 (ThermoFisher Scientific, 2020) . We provide a small command line tool mdocspoofer for the 160 generation of these metadata files for a directory containing multi-frame micrograph files of the form 161 <basename>_<count>[<tilt_angle>]_fractions.mrc, as generated by Tomography 5. This tool enables 162 use of the Warp preprocessing pipeline for data collected in Tomography 5. 163 Interfacing Dynamo and the Warp-RELION-M pipeline 164 The recommended procedure for particle picking in the Warp-RELION-M pipeline is 165 exhaustive, reference-based template matching. For initial particle pose optimisation, RELION 166 subtomogram averaging is integrated into the pipeline. Alternative approaches to particle picking, 167 pose optimisation and classification may provide advantages over this workflow. We provide 168 dynamo2m, a set of tools which create a bidirectional interface between Dynamo and the Warp169 RELION-M pipeline. dynamo2warp allows for particle position and orientation data in Dynamo to be 170 used for particle extraction in Warp and subsequent alignment in RELION and M. warp2dynamo 171 provides a route to using particles extracted in Warp within Dynamo. Because RELION modified its 172 metadata system with the release of RELION 3.1, relion_star_downgrade is provided to enable 173 extracting particles refined using RELION version 3.1 or higher in Warp 1.0.9. 174 Results and Discussion 175 Metadata handling 176 The ability to test alternative approaches and rapidly iterate is key to optimising complex data 177 analysis workflows like cryo-ET and STA. Converting between different Euler angle conventions is 178 often a pain point for those wishing to interface different software packages due to the abundance of 179 ambiguities in their interpretation (Urzhumtseva and Urzhumtsev, 2019) . The eulerangles package 180 simplifies both interfacing pieces of software which interpret Euler angles according to differing 181 conventions and making use of Euler angles for custom analyses or visualisations. 182 Implementing custom input/output functionality when attempting to work with a variety of non183 standard metadata presents a barrier to entry for those wishing to interactively explore their data or 184 implement custom analysis routines. The Python packages starfile and dynamotable remove this 185 barrier for the working with RELION and Dynamo, providing a simple interface between their 186 respective metadata systems and a wealth of data analysis infrastructure, visualisation tools and 187 educational resources (Harris et al., 2020; Paszke et al., 2019). 188 Tight integration with the scientific Python ecosystem encourages exploratory data analysis 189 and simplifies the development of any custom tooling, as is often necessary in cryo-ET. As an 190 example of the benefits derived from working within this ecosystem, we combine our starfile and 191 eulerangles packages with the packages mrcfile (Burnley et al., 2017) (for reading and writing MRC 192 format image files) and napari (Sofroniew et al., 2021) (a fast, powerful multidimensional data 193 visualisation library) to create a visualisation of particle positions, their orientations and tomograms in 194 3D (Figure 2). By combining these packages, creating such a visualisation from scratch can be 195 achieved in less than 25 lines of code (Figure 2- figure supplement 1). 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 The Dynamo - Warp-RELION-M interface

In this section, we discuss and illustrate the benefits provided by the bidirectional Dynamo Warp-RELION-M interface.

Automation of tilt-series alignment in Warp-RELION-M

Optimisation of data collection strategies has significantly increased the throughput of cryo-ET data collection in recent years, with datasets now routinely exceeding 100 tomograms. As dataset sizes increase, so does the need for accurate, automated solutions to each and every step of complex data analysis workflows. Accurate tilt-series alignment is often achieved using semiautomated procedures which quickly become tedious when faced with a large dataset. Providing the automated, robust fiducial-based tilt-series alignment procedures from Dynamo as an on-the-fly solution integrated into the Warp preprocessing pipeline using autoalign_dynamo greatly simplifies the generation of accurate 3D reconstructions for downstream data analysis. We provide no integrated solution for aligning tilt-series lacking exogenous fiducial markers, such as those from samples prepared by focused-ion beam (FIB) milling, although procedures exist in other software packages (Chen et al., 2019). As new procedures are developed, their integration into on-the-fly data processing workflows will be important for optimal use of both microscope and researcher time. Enabling geometrical approaches to subtomogram averaging in Warp-RELION-M

warp2catalogue and dynamo2warp enable the use of geometrical approaches to subtomogram averaging to users of the Warp-RELION-M pipeline (Figure 3). The catalogue system in Dynamo and the interactive 3D modelling tools in the dtmslice viewer make it easy to generate models of supporting geometries and derive particle positions and orientations for an entire dataset from minimal manual annotations (Castaño-Díez et al., 2017). Examples of easy-to-annotate supporting geometries from which positions can be easily derived are vesicles (Figure 3a), arbitrarily shaped membrane surfaces (Figure 3b), filaments and crystals. warp2catalogue simplifies the setup of a Dynamo catalogue for Warp-RELION-M users: the user annotates volumes filtered to aid visualisation, whereas particle extraction operations performed from the catalogue use the corresponding unfiltered data. This automation simplifies the use of an optimal workflow without unnecessary cognitive burden for the user.

Dynamo also provides dpktbl.subbox.tableOnTable for ‘subboxing’, deriving particle positions and orientations which are geometrically related to an existing set of positions and orientations. (Figure 3c). This procedure is often useful in cryo-ET for focussing analysis on subunits of a large complex after an initial consensus refinement (Ertel et al., 2017). dynamo2warp provides a simple mechanism for the integration of these powerful tools into the Warp-RELION-M pipeline. Providing access to alternative refinement and classification procedures

RELION implements a Bayesian approach to the refinement of one or many 3D volumes from cryo-EM data, called maximum a posteriori refinement (Scheres, 2012) . Originally designed for single particle cryo-EM data and more recently extended to work with cryo-ET data (Bharat et al., 2015), RELION provides solutions for the ‘refinement’ block of the cryo-ET pipeline. The package features an easy-to-use graphical user interface (GUI) and is already in use in a large number of cryo-EM labs. Rather than a data processing pipeline, Dynamo is a collection of powerful tools for working with subtomogram averaging data. With an emphasis on customisation, Dynamo often exposes the user to large numbers of parameters in fully featured GUIs.

By interfacing Dynamo with the Warp-RELION-M pipeline in the ‘refinement’ block, we provide Warp-RELION-M users access to highly customisable subtomogram averaging and multireference classification procedures, Principal Component Analysis (PCA)-based classification and myriad data visualisation tools and analysis tools. Dynamo users gain a route to making use of the CTF estimation and particle reconstruction tools in Warp, subtomogram averaging and classification in RELION and the multi-particle refinement procedure implemented in M.

As an example of a possible benefit of increased flexibility in subtomogram averaging workflows, we compare the options for masking between Dynamo and RELION. In Dynamo, a user may choose to provide separate masks for alignment, classification and Fourier shell correlation (FSC) to be used during an iterative alignment project. In RELION, the same mask is used for all three. Creative combination of different masks can allow for excluding a membrane from alignment for a membrane protein which would otherwise require using masks with extremely soft edges to avoid FSC artefacts. Such a soft mask would necessarily include the membrane or exclude membrane-proximal protein density.

A framework for flexible ab initio and geometrical approaches to subtomogram averaging

Each subtomogram averaging dataset is different and analysis presents its own unique challenges. The provided tools were designed to simplify cryo-ET data processing where possible and empower researchers with access to a vast array of alternative tooling where it may provide benefit. Combined, these tools enable working within a powerful, flexible framework for ab initio and geometrical approaches to subtomogram averaging, combining Warp-RELION-M with Dynamo (Figure 4). 267 268 269 270 271 272

Application to the HIV-1 CA-SP1 hexamer (EMPIAR-10164)

We illustrate the power of working within this framework on an example dataset by reprocessing a five tilt-series subset of EMPIAR-10164 containing immature HIV-1 dMACANC viruslike particles (VLPs). This dataset was contributed to the community by the Briggs group and this subset has previously been used to benchmark NovaCTF (Turoňová et al., 2017) and Warp (Tegunov and Cramer, 2019) , resulting in 3.9 Å and 3.8 Å reconstructions respectively.

Initial motion correction and initial CTF estimation were performed in Warp with respective spatiotemporal model resolutions of (1, 1, 8) and (2, 2, 1). Tilt-series were automatically aligned using autoalign_dynamo before CTF estimation and tomogram reconstruction at 10 Å/px in Warp. VLPs were annotated in Dynamo and initial estimates of positions and orientations were generated with an inter-particle distance of 45 Å, oversampled relative to the ~75 Å lattice spacing seen in the tomograms. 500 particles were extracted and averaged in Dynamo to produce a coarse template. The same data was subject to subtomogram averaging against this template in Dynamo without imposing symmetry during refinement. The resulting average contained a hexagonal lattice. To obtain an initial model, the 6-fold axis of the lattice was aligned to the z-axis and the resulting volume was symmetrised 6-fold in Dynamo. 28516 particles were extracted from annotated VLPs and aligned against the initial model for one iteration with a limited angular search range in Dynamo. The resulting particle positions and orientations, visualised in Dynamo, formed regular lattices on the surfaces of each VLP with some less regular areas. Using simple MATLAB scripts, only the 19810 particles with three or more neighbours at the expected inter-particle distance were retained for subsequent analysis. Using dynamo2m, metadata were converted to enable working in Warp-RELION-M. Particle extractions were carried out in Warp and 3D refinements in RELION 3.1, starting with local angular searches of ±15°. Resolution estimates were measured by FSC between independent reconstructions from random half-sets of the data using relion_postprocess. Particles were (i) extracted at 5 Å/px and refined (unmasked) to 10 Å resolution, (ii) extracted at 2.5 Å/px and refined (unmasked) to 5 Å resolution, (iii) extracted at 1.7 Å/px in Warp and refined (masked to include only the central hexamer) to 3.8 Å resolution. This intermediate result is indicative of the accuracy of fiducial-based alignment workflows in Dynamo and alignments in RELION. Seven rounds of multiparticle refinement were performed in M at 1.6 Å/px in a masked region containing only the central hexamer. The first 4 rounds of multi-particle refinement, optimising only for deformation parameters, yielded a 3.6 Å reconstruction. Three further rounds including optimisation of electron-optical aberration related parameters and tilt-frame alignments refined to a resolution of 3.4 Å (Figure 5). 305 306 307 308 309

It should be noted that working within this framework readily yielded a 3.4 Å reconstruction from 19810 particles in five tomograms, without use of an external reference and assuming no prior knowledge about the complex. The combination of Dynamo tools and the Warp-RELION-M pipeline, enabled by the tools presented here, rendered this ab initio, geometrical approach a smooth, efficient process. An ab initio approach is not strictly necessary for this dataset, as demonstrated by emClarity (Himes and Zhang, 2018) and Warp-RELION-M (Tegunov et al., 2021) . We however note that obtaining an initial model is often a challenging step in the first stages of a project. Thus, access to ab initio and geometrical approaches is provided as an alternative should the integrated template matching procedure prove inadequate. The final resolution of the reconstruction serves only to demonstrate the ability of M to further optimise image alignment and electron-optical parameters. Application to the E. coli chemosensory array in situ

We are currently using an optimised minicell system (Burt et al., 2020) to investigate structural bases of the chemotactic response in motile bacteria. To illustrate the advantages of working within this framework, we choose to present intermediate results from a set of 109 minicell tomograms (Figure 6).

The chemotaxis-mediating molecular machinery forms large supramolecular arrays, a highly cooperative network of core signalling units (CSUs), in the bacterial inner membrane at the cell poles. The elongated nature of the CSU, the crowded cellular milieu and increased sample thickness complicate template matching approaches with this system, increasing the false-positive rate and impeding appropriate peak extraction. Instead, we opt for a seed-oversampling approach in which initial positions and orientations are distributed on Dynamo surface models, oversampling with respect to the expected inter-particle distance of ~120 Å. Particle positions were generated from the surface model with an inter-particle distance of 30 Å using Dynamo. Particles were extracted and subject to one round of subtomogram averaging in Dynamo with local angular searches of ±30°. Constraining the angles of particles relative to initial estimates from a supporting geometry ensures 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 that particles within each array have orientational parameters coherent with our understanding of the system. Particles were extracted at 7.5 Å/px in Warp and imported into Dynamo using warp2dynamo.

A consensus refinement of these particles, centered on the three-fold symmetry axis of a p6symmetric assembly, was performed in Dynamo, allowing local deviations in out-of-plane orientation, complete search of in-plane orientation and enforcing C3 symmetry during refinement. In the resulting average, the expected 3-fold symmetric arrangement of P4 domains in the structure was not readily observed. The resulting reconstruction had a nominal resolution of 18 Å, with the FSC curve indicating the presence of significant heterogeneity. The Dynamo tool dpktbl.subbox.tableOnTable was used to derive the positions and orientations of particles centered on six inter-trimer-of-dimer axes of the array. Duplicate particles were removed using dpktbl.exclusionPerVolume prior to a consensus refinement of these particles enforcing C2 symmetry. The resulting reconstruction had a nominal resolution of 19 Å. Attempts to disentangle structural heterogeneity using multireference approaches to 3D classification in both Dynamo and RELION failed to yield sensible results at this stage.

Multi-particle refinement in M yielded improvements when performed in a masked region encompassing a large array region of 400 Å diameter, resulting in a reconstruction with a nominal resolution of 11 Å. No improvement was seen when refinement was performed on smaller regions encompassing one to four CSUs, presumably due to insufficient signal. Work is underway to improve upon this intermediate result by classification and further refinement. However, these initial results are already quite promising, showing features such as a hole at the center of the four-helical bundle of a receptor dimer, not yet seen in situ.

A platform for collaborative knowledge sharing in the cryo-ET community

Information about best practices and approaches to problems presented by cryo-ET data analysis is currently divided between the literature, mailing lists and the documentation of various pieces of software. We establish teamtomo.org as an open, collaborative platform for knowledge sharing within the growing cryo-ET community. The platform is designed from the ground-up as a collaborative endeavour with built-in mechanisms for easily integrating contributions.

Intended for sharing information about general data processing principles, problem solving approaches, short and long-form tutorials as well as providing links to relevant external material, the platform provides an overview of the field for interested parties, as well as specific practical and theoretical knowledge for those actively processing data. As an initial contribution, we provide a complete guide for the procedure we used to achieve the 3.4 Å reconstruction of the HIV-1 CA-SP1 hexamer, together with practical and theoretical tips of relevance for this and other datasets. Those wishing to share their expertise are encouraged to contribute.

The tools and resources presented here increase automation within the Warp preprocessing pipeline, simplify metadata handling and interface two complementary cryo-ET data processing solutions. The provided interface enables working within a powerful, flexible framework for geometrical and ab initio approaches to subtomogram averaging which can yield high resolution results. Our tools increase the functionality available to users of both Dynamo and Warp-RELION-M, lowering the barrier to employing more customisable workflows for cryo-ET data processing which are often key to success. Finally, we hope that the existence of a collaborative, community-driven resource for sharing cryo-ET knowledge will encourage open science, improve the ability of the growing cryo-ET community to integrate newcomers and enable their use of cryo-ET to solve lifescience questions. 387 Data availability 388 Data on immature HIV-1 dMACANC virus-like particles used for benchmarking of the pipeline 389 was taken from EMPIAR-10164. Data on E. coli minicells will be available upon publication of the final 390 reconstruction of the chemosensory array. 391 Code availability 392 Sou•rce code for all tools and resources described here is available at: 393 • autoalign_dynamo - https://github.com/alisterburt/autoalign_dynamo 394 • mdocspoofer - https://github.com/alisterburt/mdocspoofer 395 • starfile - https://github.com/alisterburt/starfile 396 • dynamotable - https://github.com/alisterburt/dynamotable 397 eulerangles - https://github.com/alisterburt/eulerangles 398 All Python packages are made available on the Python package index (PyPI). 399 Acknowledgements 400 This work has received funding from a European Union’s Horizon 2020 research and 401 innovation programme under grant agreement No 647784 to IG. We acknowledge Diamond Light 402 Source for access and support of the cryo-EM facilities at the UK's national Electron Bio-imaging 403 Centre (eBIC), funded by the Wellcome Trust, MRC and BBRSC. Cryo-ET data acquisition has been 404 supported by iNEXT, grant number 653706 (PID:2626 to IG), funded by the EU Horizon 2020 405 programme. AB is supported by a University Grenoble Alpes PhD fellowship and by a Fondation pour 406 la Recherche Medicale (FRM) fellowship. LG is supported by a PhD fellowship from Grenoble 407 Alliance for Integrated Structural and Cell Biology (GRAL, ANR-10-LABX-49-01) funded within the 408 CBH Graduate School of the University Grenoble Alpes (ANR-17-EURE-0003). 409 Competing Interests 410 The authors declare no competing interests 411 References 412 Bai X, McMullan G, Scheres SHW. 2015. How cryo-EM is revolutionizing structural biology. Trends 413 Biochem Sci 40:49–57. doi:10.1016/j.tibs.2014.10.005 414 Bartesaghi A, Lecumberry F, Sapiro G, Subramaniam S. 2012. Protein secondary structure 415 determination by constrained single-particle cryo-electron tomography. Structure 20:2003–13. 416 doi:10.1016/j.str.2012.10.016 417 Basanta B, Chowdhury S, Lander GC, Grotjahn DA. 2020. A guided approach for subtomogram 418 averaging of challenging macromolecular assemblies. J Struct Biol X 4:100041. 419 doi:10.1016/j.yjsbx.2020.100041 420 Bharat TAM, Russo CJ, Löwe J, Passmore LA, Scheres SHW. 2015. Advances in Single-Particle 421 Electron Cryomicroscopy Structure Determination applied to Sub-tomogram Averaging. 422 Structure 23:1743–1753. doi:10.1016/j.str.2015.06.026 423 Burnley T, Palmer CM, Winn M. 2017. Recent developments in the CCP-EM software suite. Acta 424 Crystallogr Sect D, Struct Biol 73:469–477. doi:10.1107/S2059798317007859 425 Burt A, Cassidy CK, Ames P, Bacia-Verloop M, Baulard M, Huard K, Luthey-Schulten Z, Desfosses A, 426 Stansfeld PJ, Margolin W, Parkinson JS, Gutsche I. 2020. Complete structure of the 427 chemosensory array core signalling unit in an E. coli minicell strain. Nat Commun 11:743. 428 429 Castaño-Díez D. 2019. Dynamo: Walkthrough on GUI based tilt series alignment. 430 https://wiki.dynamo.biozentrum.unibas.ch/w/index.php/Walkthrough_on_GUI_based_tilt_series_ 431 alignment 432 Castaño-Díez D, Kudryashev M, Arheit M, Stahlberg H. 2012. Dynamo: a flexible, user-friendly 433 development tool for subtomogram averaging of cryo-EM data in high-performance computing 434 environments. J Struct Biol 178:139–51. doi:10.1016/j.jsb.2011.12.017 435 Castaño-Díez D, Kudryashev M, Stahlberg H. 2017. Dynamo Catalogue: Geometrical tools and data 436 management for particle picking in subtomogram averaging of cryo-electron tomograms. J Struct 437 Biol 197:135–144. doi:10.1016/j.jsb.2016.06.005 438 Chen M, Bell JM, Shi X, Sun SY, Wang Z, Ludtke SJ. 2019. A complete data processing workflow for 439 cryo-ET and subtomogram averaging. Nat Methods 16:1161–1168. doi:10.1038/s41592-019440 0591-8 441 Chen M, Dai W, Sun SY, Jonasch D, He CY, Schmid MF, Chiu W, Ludtke SJ. 2017. Convolutional 442 neural networks for automated annotation of cellular cryo-electron tomograms. Nat Methods 443 14:983–985. doi:10.1038/nmeth.4405 444 Ertel KJ, Benefield D, Castaño-Diez D, Pennington JG, Horswill M, den Boon JA, Otegui MS, Ahlquist 445 P. 2017. Cryo-electron tomography reveals novel features of a viral RNA replication 446 compartment. Elife 6. doi:10.7554/eLife.25940 447 Harris CR, Millman KJ, van der Walt SJ, Gommers R, Virtanen P, Cournapeau D, Wieser E, Taylor J, 448 Berg S, Smith NJ, Kern R, Picus M, Hoyer S, van Kerkwijk MH, Brett M, Haldane A, Del Río JF, 449 Wiebe M, Peterson P, Gérard-Marchant P, Sheppard K, Reddy T, Weckesser W, Abbasi H, 450 Gohlke C, Oliphant TE. 2020. Array programming with NumPy. Nature 585:357–362. 451 doi:10.1038/s41586-020-2649-2 452 Himes BA, Zhang P. 2018. emClarity: software for high-resolution cryo-electron tomography and 453 subtomogram averaging. Nat Methods 15:955–961. doi:10.1038/s41592-018-0167-z 454 Kremer JR, Mastronarde DN, McIntosh JR. 1996. Computer visualization of three-dimensional image 455 data using IMOD. J Struct Biol 116:71–6. doi:10.1006/jsbi.1996.0013 456 Leigh KE, Navarro PP, Scaramuzza S, Chen W, Zhang Y, Castaño-Díez D, Kudryashev M. 2019. 457 Subtomogram averaging from cryo-electron tomograms. Methods Cell Biol 152:217–259. 458 doi:10.1016/bs.mcb.2019.04.003 459 Moebel E, Martinez-Sanchez A, Larivière D, Fourmentin E, Ortitz J, Baumeister W, Kervrann C. 2021. 460 Deep Learning Improves Macromolecules Localization and Identification in 3D Cellular Cryo461 Electron Tomograms. bioRxiv. doi:10.1101/2020.04.15.042747 462 Nakane T, Kotecha A, Sente A, McMullan G, Masiulis S, Brown PMGE, Grigoras IT, Malinauskaite L, 463 Malinauskas T, Miehling J, Uchański T, Yu L, Karia D, Pechnikova E V, de Jong E, Keizer J, 464 Bischoff M, McCormack J, Tiemeijer P, Hardwick SW, Chirgadze DY, Murshudov G, Aricescu AR, 465 Scheres SHW. 2020. Single-particle cryo-EM at atomic resolution. Nature 587:152–156. 466 doi:10.1038/s41586-020-2829-0 467 Paszke A, Gross S, Massa F, Lerer A, Bradbury J, Chanan G, Killeen T, Lin Z, Gimelshein N, Antiga 468 L, Desmaison A, Köpf A, Yang E, DeVito Z, Raison M, Tejani A, Chilamkurthy S, Steiner B, Fang 469 470

L, Bai J, Chintala S. 2019. PyTorch: An Imperative Style, High-Performance Deep Learning Library. 507 508