Class-specific Regions Research Articles

Distance guided generative adversarial network for explainable medical image classifications

Despite the potential benefits of data augmentation for mitigating data insufficiency, traditional augmentation methods primarily rely on prior intra-domain knowledge. On the other hand, advanced generative adversarial networks (GANs) generate inter-domain samples with limited variety. These previous methods make limited contributions to describing the decision boundaries for binary classification. In this paper, we propose a distance-guided GAN (DisGAN) that controls the variation degrees of generated samples in the hyperplane space. Specifically, we instantiate the idea of DisGAN by combining two ways. The first way is vertical distance GAN (VerDisGAN) where the inter-domain generation is conditioned on the vertical distances. The second way is horizontal distance GAN (HorDisGAN) where the intra-domain generation is conditioned on the horizontal distances. Furthermore, VerDisGAN can produce the class-specific regions by mapping the source images to the hyperplane. Experimental results show that DisGAN consistently outperforms the GAN-based augmentation methods with explainable binary classification. The proposed method can apply to different classification architectures and has the potential to extend to multi-class classification. We provide the code in https://github.com/yXiangXiong/DisGAN.

Subsequent Memory Effects in Cortical Pattern Similarity Differ by Semantic Class.

Although living and nonliving stimuli are known to rely on distinct brain regions during perception, it is largely unknown if their episodic memory encoding mechanisms differ as well. To investigate this issue, we asked participants to encode object pictures (e.g., a picture of a tiger) and to retrieve them later in response to their names (e.g., word "tiger"). For each of four semantic classes (living-animate, living-inanimate, nonliving-large, and nonliving-small), we examined differences in the similarity in activation patterns (neural pattern similarity [NPS]) for subsequently remembered versus forgotten items. Higher NPS for remembered items suggests an advantage of within-class item similarity, whereas lower NPS for remembered items indicates an advantage for item distinctiveness. We expect NPS within class-specific regions to be higher for remembered than for forgotten items. For example, the parahippocampal cortex has a well-known role in scene processing [Aminoff, E. M., Kveraga, K., & Bar, M. The role of the parahippocampal cortex in cognition. Trends in Cognitive Sciences, 17, 379-390, 2013], and the anterior temporal and inferior frontal gyrus have well-known roles in object processing [Clarke, A., & Tyler, L. K. Object-specific semantic coding in human perirhinal cortex. Journal of Neuroscience, 34, 4766-4775, 2014]. As such, we expect to see higher NPS for remembered items in these regions pertaining to scenes and objects, respectively. Consistent with this hypothesis, in fusiform, parahippocampal, and retrosplenial regions, higher NPS predicted memory for subclasses of nonliving objects, whereas in the left inferior frontal and left retrosplenial regions, lower NPS predicted memory for subclasses of living objects. Taken together, the results support the idea that subsequent memory depends on a balance of similarity and distinctiveness and demonstrate that the neural mechanisms of episodic encoding differ across semantic categories.

SATS: Self-attention transfer for continual semantic segmentation

Continually learning to segment more and more types of image regions is a desired capability for many intelligent systems. However, such continual semantic segmentation exhibits catastrophic forgetting issues similar to those of continual classification learning. Unlike the existing knowledge distillation strategies for alleviating this problem, transferring a new type of information, namely, the relationships between elements (e.g., pixels) within each image that can capture both within-class and between-class knowledge, is proposed in this study. Such information can be effectively obtained from self-attention maps in a Transformer-style segmentation model. Considering that pixels belonging to the same class in each image typically share similar visual properties, a class-specific region pooling operator is novelly applied to provide reliable relationship information for knowledge transfer. Extensive evaluations on multiple public benchmarks reveal that the proposed self-attention transfer method can effectively alleviate the catastrophic forgetting issue. Furthermore, flexible combinations of the proposed method with widely adopted strategies considerably outperform state-of-the-art solutions.

Attention-Based Fine-Grained Lightweight Architecture for Fuji Apple Maturity Classification in an Open-World Orchard Environment

Fuji apples are one of the most important and popular economic crops worldwide in the fruit industry. Nowadays, there is a huge imbalance between the urgent demand of precise automated sorting models of fruit ripeness grades due to the increasing consumption levels and the limitations of most existing methods. In this regard, this paper presents a novel CNN-based fine-grained lightweight architecture for the task of Fuji apple maturity classification (FGAL-MC). Our proposed FGAL-MC architecture has three advantages compared with related previous research works. Firstly, we established a novel Fuji apple maturity dataset. We investigated the Fuji apple’s different growth stages using image samples that were captured in open-world orchard environments, which have the benefit of being able to guide the related methods to be more suitable for the practical working environment. Secondly, because maturity grades are difficult to discriminate due to the issues of subtle expression differences, as well as the various challenging disadvantages for the unstructured surroundings, we designed our network as a fine-grained classification architecture by introducing an attention mechanism to learn class-specific regions and discrimination. Thirdly, because the number of parameters of an architecture determines the time-cost and hardware configuration to some extent, we designed our proposed architecture as a lightweight structure, which is able to be applied or promoted for actual agriculture field operations. Finally, comprehensive qualitative and quantitative experiments demonstrated that our presented method can achieve competitive results in terms of accuracy, precision, recall, F1-score, and time-cost. In addition, extensive experiments indicated our proposed method also has outstanding performance in terms of generalization ability.

Essential amino acids in the Plant-Conserved and Class-Specific Regions of cellulose synthases.

The Plant-Conserved Region (P-CR) and the Class-Specific Region (CSR) are two plant-unique sequences in the catalytic core of cellulose synthases (CESAs) for which specific functions have not been established. Here, we used site-directed mutagenesis to replace amino acids and motifs within these sequences predicted to be essential for assembly and function of CESAs. We developed an in vivo method to determine the ability of mutated CesA1 transgenes to complement an Arabidopsis (Arabidopsis thaliana) temperature-sensitive root-swelling1 (rsw1) mutant. Replacement of a Cys residue in the CSR, which blocks dimerization in vitro, rendered the AtCesA1 transgene unable to complement the rsw1 mutation. Examination of the CSR sequences from 33 diverse angiosperm species showed domains of high-sequence conservation in a class-specific manner but with variation in the degrees of disorder, indicating a nonredundant role of the CSR structures in different CESA isoform classes. The Cys residue essential for dimerization was not always located in domains of intrinsic disorder. Expression of AtCesA1 transgene constructs, in which Pro417 and Arg453 were substituted for Ala or Lys in the coiled-coil of the P-CR, were also unable to complement the rsw1 mutation. Despite an expected role for Arg457 in trimerization of CESA proteins, AtCesA1 transgenes with Arg457Ala mutations were able to fully restore the wild-type phenotype in rsw1. Our data support that Cys662 within the CSR and Pro417 and Arg453 within the P-CR of Arabidopsis CESA1 are essential residues for functional synthase complex formation, but our data do not support a specific role for Arg457 in trimerization in native CESA complexes.

Fault detection method for energy measurement systems equipped with a Rogowski coil using the coil's response to a unit voltage jump and a fully convolutional neural network

The article presents a method of assessing the condition of a measurement system equipped with a Rogowski coil using the response of the coil to the unit voltage jump in the secondary circuit. The method is based on a direct sensor-microcontroller interface and has been tested on the STM32F745 microcontroller. Unlike traditional direct sensor-microcontroller methods described in literature a Fully Convolutional Neural Network (FCN) is used to extract signal features and estimate the state of the system. The microcontroller is responsible for capturing the coil responses, which are used by the FCN for time series classification. This method allows creating smart sensor with self-testing and identification capabilities. The Class Activation Map (CAM) is used to define class specific contribution regions and verify the performance of the FCN network. The proposed framework is suitable for remote assessment of the system condition in high voltage areas where Rogowski coils are used. Because of the presence of voltages dangerous for humans and the frequent inability to switch off the voltage in a power facility, this method significantly speeds up the location of damage in the measurement system.

Bioinformatic Analyses of Peroxiredoxins and RF-Prx: A Random Forest-Based Predictor and Classifier for Prxs.

Peroxiredoxins (Prxs) are a protein superfamily, present in all organisms, that play a critical role in protecting cellular macromolecules from oxidative damage but also regulate intracellular and intercellular signaling processes involving redox-regulated proteins and pathways. Bioinformatic approaches using computational tools that focus on active site-proximal sequence fragments (known as active site signatures) and iterative clustering and searching methods (referred to as TuLIP and MISST) have recently enabled the recognition of over 38,000 peroxiredoxins, as well as their classification into six functionally relevant groups. With these data providing so many examples of Prxs in each class, machine learning approaches offer an opportunity to extract additional information about features characteristic of these protein groups.In this study, we developed a novel computational method named "RF-Prx" based on a random forest (RF) approach integrated with K-space amino acid pairs (KSAAP) to identify peroxiredoxins and classify them into one of six subgroups. Our process performed in a superior manner compared to other machine learning classifiers. Thus the RF approach integrated with K-space amino acid pairs enabled the detection of class-specific conserved sequences outside the known functional centers and with potential importance. For example, drugs designed to target Prx proteins would likely suffer from cross-reactivity among distinct Prxs if targeted to conserved active sites, but this may be avoidable if remote, class-specific regions could be targeted instead.

A Visually Interpretable Deep Learning Framework for Histopathological Image-Based Skin Cancer Diagnosis.

Owing to the high incidence rate and the severe impact of skin cancer, the precise diagnosis of malignant skin tumors is a significant goal, especially considering treatment is normally effective if the tumor is detected early. Limited published histopathological image sets and the lack of an intuitive correspondence between the features of lesion areas and a certain type of skin cancer pose a challenge to the establishment of high-quality and interpretable computer-aided diagnostic (CAD) systems. To solve this problem, a light-weight attention mechanism-based deep learning framework, namely, DRANet, is proposed to differentiate 11 types of skin diseases based on a real histopathological image set collected by us during the last 10 years. The CAD system can output not only the name of a certain disease but also a visualized diagnostic report showing possible areas related to the disease. The experimental results demonstrate that the DRANet obtains significantly better performance than baseline models (i.e., InceptionV3, ResNet50, VGG16, and VGG19) with comparable parameter size and competitive accuracy with fewer model parameters. Visualized results produced by the hidden layers of the DRANet actually highlight part of the class-specific regions of diagnostic points and are valuable for decision making in the diagnosis of skin diseases.

Improved Semantic Segmentation of Tuberculosis-Consistent Findings in Chest X-rays Using Augmented Training of Modality-Specific U-Net Models with Weak Localizations.

Deep learning (DL) has drawn tremendous attention for object localization and recognition in both natural and medical images. U-Net segmentation models have demonstrated superior performance compared to conventional hand-crafted feature-based methods. Medical image modality-specific DL models are better at transferring domain knowledge to a relevant target task than those pretrained on stock photography images. This character helps improve model adaptation, generalization, and class-specific region of interest (ROI) localization. In this study, we train chest X-ray (CXR) modality-specific U-Nets and other state-of-the-art U-Net models for semantic segmentation of tuberculosis (TB)-consistent findings. Automated segmentation of such manifestations could help radiologists reduce errors and supplement decision-making while improving patient care and productivity. Our approach uses the publicly available TBX11K CXR dataset with weak TB annotations, typically provided as bounding boxes, to train a set of U-Net models. Next, we improve the results by augmenting the training data with weak localization, postprocessed into an ROI mask, from a DL classifier trained to classify CXRs as showing normal lungs or suspected TB manifestations. Test data are individually derived from the TBX11K CXR training distribution and other cross-institutional collections, including the Shenzhen TB and Montgomery TB CXR datasets. We observe that our augmented training strategy helped the CXR modality-specific U-Net models achieve superior performance with test data derived from the TBX11K CXR training distribution and cross-institutional collections (p < 0.05). We believe that this is the first study to i) use CXR modality-specific U-Nets for semantic segmentation of TB-consistent ROIs and ii) evaluate the segmentation performance while augmenting the training data with weak TB-consistent localizations.

Structure of Arabidopsis CESA3 catalytic domain with its substrate UDP-glucose provides insight into the mechanism of cellulose synthesis

Cellulose is synthesized by cellulose synthases (CESAs) from the glycosyltransferase GT-2 family. In plants, the CESAs form a six-lobed rosette-shaped CESA complex (CSC). Here we report crystal structures of the catalytic domain of Arabidopsis thaliana CESA3 (AtCESA3CatD) in both apo and uridine diphosphate (UDP)-glucose (UDP-Glc)-bound forms. AtCESA3CatD has an overall GT-A fold core domain sandwiched between a plant-conserved region (P-CR) and a class-specific region (C-SR). By superimposing the structure of AtCESA3CatD onto the bacterial cellulose synthase BcsA, we found that the coordination of the UDP-Glc differs, indicating different substrate coordination during cellulose synthesis in plants and bacteria. Moreover, structural analyses revealed that AtCESA3CatD can form a homodimer mainly via interactions between specific beta strands. We confirmed the importance of specific amino acids on these strands for homodimerization through yeast and in planta assays using point-mutated full-length AtCESA3. Our work provides molecular insights into how the substrate UDP-Glc is coordinated in the CESAs and how the CESAs might dimerize to eventually assemble into CSCs in plants.

The molecular basis of plant cellulose synthase complex organisation and assembly.

The material properties of cellulose are heavily influenced by the organisation of β-1,4-glucan chains into a microfibril. It is likely that the structure of this microfibril is determined by the spatial arrangement of catalytic cellulose synthase (CESA) proteins within the cellulose synthase complex (CSC). In land plants, CESA proteins form a large complex composed of a hexamer of trimeric lobes termed the rosette. Each rosette synthesises a single microfibril likely composed of 18 glucan chains. In this review, the biochemical events leading to plant CESA protein assembly into the rosette are explored. The protein interfaces responsible for CESA trimerization are formed by regions that define rosette-forming CESA proteins. As a consequence, these regions are absent from the ancestral bacterial cellulose synthases (BcsAs) that do not form rosettes. CSC assembly occurs within the context of the endomembrane system, however the site of CESA assembly into trimers and rosettes is not determined. Both the N-Terminal Domain and Class Specific Region of CESA proteins are intrinsically disordered and contain all of the identified phosphorylation sites, making both regions candidates as sites for protein-protein interactions and inter-lobe interface formation. We propose a sequential assembly model, whereby CESA proteins form stable trimers shortly after native folding, followed by sequential recruitment of lobes into a rosette, possibly assisted by Golgi-localised STELLO proteins. A comprehensive understanding of CESA assembly into the CSC will enable directed engineering of CESA protein spatial arrangements, allowing changes in cellulose crystal packing that alter its material properties.

Ensemble model with cascade attention mechanism for high-resolution remote sensing image scene classification.

Scene classification of high-resolution remote sensing images is a fundamental task of earth observation. And numerous methods have been proposed to achieve this. However, these models are inadequate as the number of labelled training data limits them. Most of the existing methods entirely rely on global information, while regions with class-specific ground objects determine the categories of high-resolution remote sensing images. An ensemble model with a cascade attention mechanism, which consists of two kinds of the convolutional neural network, is proposed to address these issues. To improve the generality of the feature extractor, each branch is trained on different large datasets to enrich the prior knowledge. Moreover, to force the model to focus on the most class-specific region in each high-resolution remote sensing image, a cascade attention mechanism is proposed to combine the branches and capture the most discriminative information. By experiments on four benchmark datasets, OPTIMAL-31, UC Merced Land-Use Dataset, Aerial Image Dataset and NWPU-RESISC45, the proposed end-to-end model cascade attention-based double branches model in this paper achieves state-of-the-art performance on each benchmark dataset.

Endosidin20 Targets the Cellulose Synthase Catalytic Domain to Inhibit Cellulose Biosynthesis

Plant cellulose is synthesized by rosette-structured cellulose synthase (CESA) complexes (CSCs). Each CSC is composed of multiple subunits of CESAs representing three different isoforms. Individual CESA proteins contain conserved catalytic domains for catalyzing cellulose synthesis, other domains such as plant-conserved sequences, and class-specific regions that are thought to facilitate complex assembly and CSC trafficking. Because of the current lack of atomic-resolution structures for plant CSCs or CESAs, the molecular mechanism through which CESA catalyzes cellulose synthesis and whether its catalytic activity influences efficient CSC transport at the subcellular level remain unknown. Here, by performing chemical genetic analyses, biochemical assays, structural modeling, and molecular docking, we demonstrate that Endosidin20 (ES20) targets the catalytic site of CESA6 in Arabidopsis (Arabidopsis thaliana). Chemical genetic analysis revealed important amino acids that potentially participate in the catalytic activity of plant CESA6, in addition to previously identified conserved motifs across kingdoms. Using high spatiotemporal resolution live cell imaging, we found that inhibiting the catalytic activity of CESA6 by ES20 treatment reduced the efficiency of CSC transport to the plasma membrane. Our results demonstrate that ES20 is a chemical inhibitor of CESA activity and trafficking that represents a powerful tool for studying cellulose synthesis in plants.

Biochemical and physiological flexibility accompanies reduced cellulose biosynthesis in Brachypodium cesa1 S830N.

Here, we present a study into the mechanisms of primary cell wall cellulose formation in grasses, using the model cereal grass Brachypodium distachyon. The exon found adjacent to the BdCESA1 glycosyltransferase QXXRW motif was targeted using Targeting Induced Local Lesions in Genomes (TILLING) and sequencing candidate amplicons in multiple parallel reactions (SCAMPRing) leading to the identification of the Bdcesa1S830N allele. Plants carrying this missense mutation exhibited a significant reduction in crystalline cellulose content in tissues that rely on the primary cell wall for biomechanical support. However, Bdcesa1S830N plants failed to exhibit the predicted reduction in plant height. In a mechanism unavailable to eudicotyledons, B. distachyon plants homozygous for the Bdcesa1S830N allele appear to overcome the loss of internode expansion anatomically by increasing the number of nodes along the stem. Stem biomechanics were resultantly compromised in Bdcesa1S830N. The Bdcesa1S830N missense mutation did not interfere with BdCESA1 gene expression. However, molecular dynamic simulations of the CELLULOSE SYNTHASE A (CESA) structure with modelled membrane interactions illustrated that Bdcesa1S830N exhibited structural changes in the translated gene product responsible for reduced cellulose biosynthesis. Molecular dynamic simulations showed that substituting S830N resulted in a stabilizing shift in the flexibility of the class specific region arm of the core catalytic domain of CESA, revealing the importance of this motion to protein function.

Cellulose synthase 'class specific regions' are intrinsically disordered and functionally undifferentiated.

Cellulose synthases (CESAs) are glycosyltransferases that catalyze formation of cellulose microfibrils in plant cell walls. Seed plant CESA isoforms cluster in six phylogenetic clades, whose non-interchangeable members play distinct roles within cellulose synthesis complexes (CSCs). A 'class specific region' (CSR), with higher sequence similarity within versus between functional CESA classes, has been suggested to contribute to specific activities or interactions of different isoforms. We investigated CESA isoform specificity in the moss, Physcomitrella patens (Hedw.) B. S. G. to gain evolutionary insights into CESA structure/function relationships. Like seed plants, P. patens has oligomeric rosette-type CSCs, but the PpCESAs diverged independently and form a separate CESA clade. We showed that P. patens has two functionally distinct CESAs classes, based on the ability to complement the gametophore-negative phenotype of a ppcesa5 knockout line. Thus, non-interchangeable CESA classes evolved separately in mosses and seed plants. However, testing of chimeric moss CESA genes for complementation demonstrated that functional class-specificity is not determined by the CSR. Sequence analysis and computational modeling showed that the CSR is intrinsically disordered and contains predicted molecular recognition features, consistent with a possible role in CESA oligomerization and explaining the evolution of class-specific sequences without selection for class-specific function.

Evolution, gene expression profiling and 3D modeling of CSLD proteins in cotton

BackgroundAmong CESA-like gene superfamily, the cellulose synthase-like D (CSLD) genes are most similar to cellulose synthase genes and have been reported to be involved in tip-growing cell and stem development. However, there has been no genome-wide characterization of this gene subfamily in cotton. We thus sought to analyze the evolution and functional characterization of CSLD proteins in cotton based on fully sequenced cotton genomes.ResultsA total of 23 full-length CSLD proteins were identified in Gossypium raimondii, Gossypium arboreum and Gossypium hirsutum. The phylogenetic tree divided the CSLD proteins into five clades with strong support: CSLD1, CSLD2/3, CSLD4, CSLD5 and CSLD6. The total expression of GhCSLD genes was the highest in androecium & gynoecium (mostly contributed by CSLD1 and CSLD4) compared with other CSL genes. CSLD1 and CSLD4 were only highly expressed in androecium & gynoecium (A&G), and showed tissue-specific expression. The total expression of CSLD2/3, 5 and 6 was highest in the specific tissues. These results suggest that CSLD genes showed the different pattern of expression. Cotton CSLD proteins were subjected to different evolutionary pressures, and the CSLD1 and CSLD4 proteins exhibited episodic and long-term shift positive selection. The predicted three-dimensional structure of GrCSLD1 suggested that GrCSLD1 belongs to glycosyltransferase family 2. The amino acid residues under positive selection in the CSLD1 lineage are positioned in a region adjacent to the class-specific region (CSR), β1-strand and transmembrane helices (TMHs) in the GrCSLD1structure.ConclusionOur results characterized the CSLD proteins by an integrated approach containing phylogeny, transcriptional profiling and 3D modeling. The study added to the understanding about the importance of the CSLD family and provide a useful reference for selecting candidate genes and their associations with the biosynthesis of the cell wall in cotton.

Discovering characteristic landmarks on ancient coins using convolutional networks

In this paper, we propose a novel method to find characteristic landmarks on ancient Roman imperial coins using deep convolutional neural network models (CNNs). We formulate an optimization problem to discover class-specific regions while guaranteeing specific controlled loss of accuracy. Analysis on visualization of the discovered region confirms that not only can the proposed method successfully find a set of characteristic regions per class, but also the discovered region is consistent with human expert annotations. We also propose a new framework to recognize the Roman coins which exploits hierarchical structure of the ancient Roman coins using the state-of-the-art classification power of the CNNs adopted to a new task of coin classification. Experimental results show that the proposed framework is able to effectively recognize the ancient Roman coins. For this research, we have collected a new Roman coin dataset where all coins are annotated and consist of observe (head) and reverse (tail) images.

Class-specific image representation for image classification using multiple scale-invariant region detectors

We propose a new class-specific image representation for image classification using multiple region detectors. The new representation is designed to solve the problem of increasing variation in object location and size within images of a class, for which traditional spatial pyramid matching shows limited classification accuracy. We propose a new region-division method that divides the image region into two class-specific regions, called class-specific region-of-interest (C-ROI) and focal region (FR). Using multiple region detectors and appropriate mixing of their responses avoids the problem of selecting a region detector that gives the best classification accuracy for a given image class, and thereby yields better results than using only one region detector. Several scale-invariant region detectors are used to obtain C-ROI and FR by considering their importance over a given image class. In experiments using several well-known datasets, the proposed method improved the accuracy and achieved results that were better than or comparable to those achieved by the related methods.

A Structural Study of CESA1 Catalytic Domain of Arabidopsis Cellulose Synthesis Complex: Evidence for CESA Trimers.

A cellulose synthesis complex with a "rosette" shape is responsible for synthesis of cellulose chains and their assembly into microfibrils within the cell walls of land plants and their charophyte algal progenitors. The number of cellulose synthase proteins in this large multisubunit transmembrane protein complex and the number of cellulose chains in a microfibril have been debated for many years. This work reports a low resolution structure of the catalytic domain of CESA1 from Arabidopsis (Arabidopsis thaliana; AtCESA1CatD) determined by small-angle scattering techniques and provides the first experimental evidence for the self-assembly of CESA into a stable trimer in solution. The catalytic domain was overexpressed in Escherichia coli, and using a two-step procedure, it was possible to isolate monomeric and trimeric forms of AtCESA1CatD. The conformation of monomeric and trimeric AtCESA1CatD proteins were studied using small-angle neutron scattering and small-angle x-ray scattering. A series of AtCESA1CatD trimer computational models were compared with the small-angle x-ray scattering trimer profile to explore the possible arrangement of the monomers in the trimers. Several candidate trimers were identified with monomers oriented such that the newly synthesized cellulose chains project toward the cell membrane. In these models, the class-specific region is found at the periphery of the complex, and the plant-conserved region forms the base of the trimer. This study strongly supports the "hexamer of trimers" model for the rosette cellulose synthesis complex that synthesizes an 18-chain cellulose microfibril as its fundamental product.

Prediction of the structures of the plant-specific regions of vascular plant cellulose synthases and correlated functional analysis

Seed plants express cellulose synthase (CESA) protein isoforms with non-redundant functions, but how the isoforms function differently is unknown. Compared to bacterial cellulose synthases, CESAs have two insertions in the large cytosolic loop: the relatively well-conserved Plant Conserved Region (P-CR) and a Class Specific Region (CSR) that varies between CESAs. Absent any atomic structure of a plant CESA, we used ab initio protein structure prediction and molecular modeling to explore how these plant-specific regions may modulate CESA function. We modeled P-CR and CSR peptides from Arabidopsis thaliana CESAs representing the six clades of seed plant CESAs. As expected, the predicted wild type P-CR structures were similar. Modeling of the mutant P-CR of Atcesa8 R362K (fra6) suggested that changes in local structural stability and surface electrostatics may cause the mutant phenotype. Among CSRs within CESAs required for primary wall cellulose synthesis, the amino sequence and the modeled arrangement of helices was most similar in AtCESA1 and AtCESA3. Genetic complementation of known Arabidopsis mutants showed that the CSRs of AtCESA1 and AtCESA3 can function interchangeably in vivo. Analysis of protein surface electrostatics led to ideas about how the surface charges on CSRs may mediate protein–protein interactions. Refined modeling of the P-CR and CSR regions of GhCESA1 from cotton modified their tertiary structures, spatial relationships to the catalytic domain, and preliminary predictions about CESA oligomer formation. Cumulatively, the results provide structural clues about the function of plant-specific regions of CESA.