Interfacing Spinal Motor Units in Non-Human Primates Identifies a Principal Neural Component for Force Control Constrained by the Size Principle

ABSTRACTMotor units convert the last neural code of movement into muscle forces. The classic view of motor unit control is that the central nervous system sends common synaptic inputs to motoneuron pools and that motoneurons respond in an orderly fashion dictated by the size principle. This view however is in contrast with the large number of dimensions observed in motor cortex which may allow individual and flexible control of motor units. Evidence for flexible control of motor units may be obtained by tracking motor units longitudinally during the performance of tasks with some level of behavioural variability. Here we identified and tracked populations of motor units in the brachioradialis muscle of two macaque monkeys during ten sessions spanning over one month during high force isometric contractions with a broad range of rate of force development (1.8 – 38.6 N·m·s-1). During the same sessions we recorded intramuscular EMG signals from 16 arm muscles of both limbs and elicited the full recruitment through neural stimulation of the median and deep radial nerves. We found a very stable recruitment order and discharge characteristics of the motor units over sessions and contraction trials. The small deviations from orderly recruitment were observed between motor units with close recruitment thresholds, and only during high rate of force development. Moreover, we also found that one component explained more than ~50% of the motor unit discharge rate variance, and that the remaining components could be described as a time-shifted version of the first, as it could be predicted from the interplay between the size principle of recruitment and one common input. In conclusion, our results show that motoneurons recruitment is determined by the interplay of the size principle and common input and that this recruitment scheme is not violated over time nor by the speed of the contractions.

snpQT: flexible, reproducible, and comprehensive quality control and imputation of genomic data

Quality control of genomic data is an essential but complicated multi-step procedure, often requiring separate installation and expert familiarity with a combination of different bioinformatics tools. Software incompatibilities, and inconsistencies across computing environments, are recurrent challenges, leading to poor reproducibility. Existing semi-automated or automated solutions lack comprehensive quality checks, flexible workflow architecture, and user control. To address these challenges, we have developed snpQT: a scalable, stand-alone software pipeline using nextflow and BioContainers, for comprehensive, reproducible and interactive quality control of human genomic data. snpQT offers some 36 discrete quality filters or correction steps in a complete standardised pipeline, producing graphical reports to demonstrate the state of data before and after each quality control procedure. This includes human genome build conversion, population stratification against data from the 1,000 Genomes Project, automated population outlier removal, and built-in imputation with its own pre- and post- quality controls. Common input formats are used, and a synthetic dataset and comprehensive online tutorial are provided for testing, educational purposes, and demonstration. The snpQT pipeline is designed to run with minimal user input and coding experience; quality control steps are implemented with numerous user-modifiable thresholds, and workflows can be flexibly combined in custom combinations. snpQT is open source and freely available at https://github.com/nebfield/snpQT. A comprehensive online tutorial and installation guide is provided through to GWAS (https://snpqt.readthedocs.io/en/latest/), introducing snpQT using a synthetic demonstration dataset and a real-world Amyotrophic Lateral Sclerosis SNP-array dataset.

On the interpretation of deep learning models in bearing vibration diagnostics

In recent years, data-driven techniques such as deep learning (DL), have been widely represented in the literature in the field of bearing vibration condition monitoring. While these approaches achieve excellent performance in detecting bearing faults on controlled laboratory datasets, there is little information available on their applicability to more realistic working conditions. One challenge of these data-driven approaches is that they can learn non-classical features unrelated to the physical defect, making their generalizability debatable. To overcome the challenge of generalizability in DL models, we aim to first understand the underlying representation that the network uses to classify different bearing defects. Having an interpretable DL model may give us hints on how to increase its applicability by, e.g., data augmentation, changing input representations or adapting model architectures. To benefit from advances in interpretability in DL methods from computer vision, we first transform the vibration signal into an image. We evaluate a common input transformation, namely the spectrogram. Subsequently, the representations that the network has learnt are evaluated. We use the Grad-CAM algorithm together with signal modifications to evaluate which parts of the input signal contribute to class attribution. Our results show that the network learns signal features related to the transfer path, the physical properties of the test setup, rather than picking up classical features having a physical relation with the defect. Given that a transfer path is very machine specific, this could be an explanation for the lack of scalability of DL methods. To improve the generalizability of DL methods on bearing vibration analysis, the competing dominant machine specific features should be eliminated from the input representation. These results highlight the importance of combining domain expertise with data-driven approaches.

Artificial Intelligence—A Tool for Risk Assessment of Delayed-Graft Function in Kidney Transplant

Delayed-graft function (DGF) might be responsible for shorter graft survival. Therefore, a clinical tool predicting its occurrence is vital for the risk assessment of transplant outcomes. In a single-center study, we conducted data mining and machine learning experiments, resulting in DGF predictive models based on random forest classifiers (RF) and an artificial neural network called multi-layer perceptron (MLP). All designed models had four common input parameters, determining the best accuracy and discriminant ability: donor’s eGFR, recipient’s BMI, donor’s BMI, and recipient–donor weight difference. RF and MLP designs, using these parameters, achieved an accuracy of 84.38% and an area under curve (AUC) 0.84. The model additionally implementing a donor’s age, gender, and Kidney Donor Profile Index (KDPI) accomplished an accuracy of 93.75% and an AUC of 0.91. The other configuration with the estimated post-transplant survival (EPTS) and the kidney donor risk profile (KDRI) achieved an accuracy of 93.75% and an AUC of 0.92. Using machine learning, we were able to assess the risk of DGF in recipients after kidney transplant from a deceased donor. Our solution is scalable and can be improved during subsequent transplants. Based on the new data, the models can achieve better outcomes.

Functional connectivity networks of common synaptic inputs to motor neurons reveal neural spinal synergies during a multi-joint task

Movements are reportedly controlled through the combination of synergies that generate specific motor outputs by imposing an activation pattern on a group of muscles. To date, the smallest unit of analysis has been the muscle through the measurement of its activation. However, the muscle is not the lowest neural level of movement control. In this human study, we identified the common synaptic inputs received by motor neurons during an isometric multi-joint task. We decoded the spiking activities of dozens of spinal motor neurons innervating six lower limb muscles in 10 participants. Furthermore, we analyzed these activities by identifying their common low-frequency components, from which networks of common synaptic inputs to the motor neurons were derived. The vast majority of the identified motor neurons shared common inputs with other motor neuron(s). In addition, groups of motor neurons were partly decoupled from their innervated muscle, such that motor neurons innervating the same muscle did not necessarily receive common inputs. Conversely, some motor neurons from different muscles, including distant muscles, received common inputs. Our results provide evidence of a synergistic control of a multi-joint motor task at the spinal motor neuron level. Moreover, we showed that common input to motor neurons is an essential feature of the neural control of movement. We conclude that the central nervous system controls flexible groups of motor neurons by distributing common inputs to substantially reduce the dimensionality of movement control.

The control and training of single motor units in isometric tasks are constrained by a common synaptic input signal

Recent developments in neural interfaces enable the real-time and non-invasive tracking of motor neuron spiking activity. Such novel interfaces provide a promising basis for human motor augmentation by extracting potential high-dimensional control signals directly from the human nervous system. However, it is unclear how flexibly humans can control the activity of individual motor neurones to effectively increase the number of degrees-of-freedom available to coordinate multiple effectors simultaneously. Here, we provided human subjects (N=7) with real-time feedback on the discharge patterns of pairs of motor units (MUs) innervating a single muscle (tibialis anterior) and encouraged them to independently control the MUs by tracking targets in a 2D space. Subjects learned control strategies to achieve the target-tracking task for various combinations of MUs. These strategies rarely corresponded to a volitional control of independent input signals to individual MUs. Conversely, MU activation was consistent with a common input to the MU pair, while individual activation of the MUs in the pair was predominantly achieved by alterations in de-recruitment order that could be explained with history-dependent changes in motor neuron excitability. These results suggest that flexible MU control based on independent synaptic inputs to single MUs is not a simple to learn control strategy.

Ultrafast population coding and axo-somatic compartmentalization

Populations of cortical neurons respond to common input within a millisecond. Morphological features and active ion channel properties were suggested to contribute to this astonishing processing speed. Here we report an exhaustive study of ultrafast population coding for varying axon initial segment (AIS) location, soma size, and axonal current properties. In particular, we studied their impact on two experimentally observed features 1) precise action potential timing, manifested in a wide-bandwidth dynamic gain, and 2) high-frequency boost under slowly fluctuating correlated input. While the density of axonal channels and their distance from the soma had a very small impact on bandwidth, it could be moderately improved by increasing soma size. When the voltage sensitivity of axonal currents was increased we observed ultrafast coding and high-frequency boost. We conclude that these computationally relevant features are strongly dependent on axonal ion channels' voltage sensitivity, but not their number or exact location. We point out that ion channel properties, unlike dendrite size, can undergo rapid physiological modification, suggesting that the temporal accuracy of neuronal population encoding could be dynamically regulated. Our results are in line with recent experimental findings in AIS pathologies and establish a framework to study structure-function relations in AIS molecular design.

snpQT: flexible, reproducible, and comprehensive quality control and imputation of genomic data

Quality control of genomic data is an essential but complicated multi-step procedure, often requiring separate installation and expert familiarity with a combination of different bioinformatics tools. Dependency hell and reproducibility are recurrent challenges. Existing semi-automated or automated solutions lack comprehensive quality checks, flexible workflow architecture, and user control. To address these challenges, we have developed snpQT: a scalable, stand-alone software pipeline using nextflow and BioContainers, for comprehensive, reproducible and interactive quality control of human genomic data. snpQT offers some 36 discrete quality filters or correction steps in a complete standardised pipeline, producing graphical reports to demonstrate the state of data before and after each quality control procedure. This includes human genome build conversion, population stratification against data from the 1,000 Genomes Project, automated population outlier removal, and built-in imputation with its own pre- and post- quality controls. Common input formats are used, and a synthetic dataset and comprehensive online tutorial are provided for testing, educational purposes, and demonstration. The snpQT pipeline is designed to run with minimal user input and coding experience; quality control steps are implemented with default thresholds which can be modified by the user, and workflows can be flexibly combined in custom combinations. snpQT is open source and freely available at https://github.com/nebfield/snpQT. A comprehensive online tutorial and installation guide is provided through to GWAS (https://snpqt.readthedocs.io/en/latest/), introducing snpQT using a synthetic demonstration dataset and a real-world Amyotrophic Lateral Sclerosis SNP-array dataset.

Straintables: An application that extracts sequences from genome assemblies and generates dissimilarity matrices

Background and Objectives: The dissimilarity matrix (DM) is an important component of phylogenetic analysis, and many software packages exist to build and show DMs. However, as the common input for this type of software are sequences in FASTA file format, the process of extracting and aligning each set of sequences to produce a big number of matrices can be laborious. Additionally, existing software does not facilitate the comparison of clusters of similarity across several DMs built for the same group of individuals, using different genomic regions. To address our requirements of such a tool, we designed Straintables to extract specific genomic region sequences from a group of intraspecies genomic assemblies, using extracted sequences to build dissimilarity matrices. Methods: A Python module with executable scripts was developed for a study on genetic diversity across strains of Toxoplasma gondii, being a general purpose system for DM calculation and visualization for preliminary phylogenetic studies. For automatic region sequence extraction from genomic assemblies we assembled a system that designs virtual primers using reference sequences located at genomic annotations, then matches those primers on genome files by using regex patterns. Extracted sequences are then aligned using Clustal Omega and compared to generate matrices. Results: Using this software saves the user from manual preparation and alignment of the sequences, a process that can be laborious when a large number of assemblies or regions are involved. The automatic sequence extraction process can be checked against BLAST results using the extracted sequence as queries, where correct results were observed for same-species pools for various organisms. The package also contains a matrix visualization tool focused on cluster visualization, capable of drawing matrices into image files with custom settings, and features methods of reordering matrices to facilitate the comparison of clustering patterns across two or more matrices. Conclusion: Straintables may replace and extend the functionality of existing matrix-oriented phylogenetic software, featuring automatic region extraction from genomic assemblies and enhanced matrix visualization capabilities emphasizing cluster identification. This module is open source, available at GitHub (https://github.com/Gab0/straintables) under a MIT license and also as a PIPY package.

Local field potentials in the parietal reach region reveal mechanisms of bimanual coordination

Primates use their arms in complex ways that frequently require coordination between the two arms. Yet the planning of bimanual movements has not been well-studied. We recorded spikes and local field potentials (LFP) from the parietal reach region (PRR) in both hemispheres simultaneously while monkeys planned and executed unimanual and bimanual reaches. From analyses of interhemispheric LFP-LFP and spike-LFP coherence, we found that task-specific information is shared across hemispheres in a frequency-specific manner. This shared information could arise from common input or from direct communication. The population average unit activity in PRR, representing PRR output, encodes only planned contralateral arm movements while beta-band LFP power, a putative PRR input, reflects the pattern of planned bimanual movement. A parsimonious interpretation of these data is that PRR integrates information about the movement of the left and right limbs, perhaps in service of bimanual coordination.