Sunkara, Vikram
Refine
Document Type
- Article (23)
- ZIB-Report (4)
- In Proceedings (2)
Keywords
- Chemical Master Equations (1)
- Dimension Reduction (1)
- Markov Chains (1)
- Opioid, Ligand-Receptor Interaction, Binding Kinetics, Molecular Dynamics, Metadynamics, SQRA (1)
- mesoscale spreading process (1)
- mesoscale spreading process, network inference, time-evolving network, romanization spreading, scarce data (1)
- network inference (1)
- romanization spreading (1)
- scarce data (1)
- time-evolving network (1)
Institute
- Modeling and Simulation of Complex Processes (16)
- Numerical Mathematics (14)
- Visual and Data-centric Computing (12)
- Mathematics for Life and Materials Science (2)
- Computational Molecular Design (1)
- Computational Systems Biology (1)
- Distributed Algorithms and Supercomputing (1)
- Mathematical Algorithmic Intelligence (1)
5q-spinal muscular atrophy (SMA) is a neuromuscular disorder (NMD) that has become one of the first 5% treatable rare diseases. The efficacy of new SMA therapies is creating a dynamic SMA patient landscape, where disease progression and scoliosis development play a central role, however, remain difficult to anticipate. New approaches to anticipate disease progression and associated sequelae will be needed to continuously provide these patients the best standard of care. Here we developed an interpretable machine learning (ML) model that can function as an assistive tool in the anticipation of SMA-associated scoliosis based on disease progression markers. We collected longitudinal data from 86 genetically confirmed SMA patients. We selected six features routinely assessed over time to train a random forest classifier. The model achieved a mean accuracy of 0.77 (SD 0.2) and an average ROC AUC of 0.85 (SD 0.17). For class 1 ‘scoliosis’ the average precision was 0.84 (SD 0.11), recall 0.89 (SD 0.22), F1-score of 0.85 (SD 0.17), respectively. Our trained model could predict scoliosis using selected disease progression markers and was consistent with the radiological measurements. During post validation, the model could predict scoliosis in patients who were unseen during training. We also demonstrate that rare disease data sets can be wrangled to build predictive ML models. Interpretable ML models can function as assistive tools in a changing disease landscape and have the potential to democratize expertise that is otherwise clustered at specialized centers.
Modern Deep Neural Networks are getting wider and deeper in their architecture design. However, with an increasing number of parameters the decision mechanisms becomes more opaque. Therefore, there is a need for understanding the structures arising in the hidden layers of deep neural networks.
In this work, we present a new mathematical framework for describing the canonical polyhedral decomposition in the input space, and in addition, we introduce the notions of collapsing- and preserving patches, pertinent to understanding the forward map and the activation space they induce.
The activation space can be seen as the output of a layer and, in the particular case of ReLU activations, we prove that this output has the structure of a polyhedral complex.
Background: Despite recent advances in cellular cryo-electron tomography (CET), developing automated tools for macromolecule identification in submolecular resolution remains challenging due to the lack of annotated data and high structural complexities. To date, the extent of the deep learning methods constructed for this problem is limited to conventional Convolutional Neural Networks (CNNs). Identifying macromolecules of different types and sizes is a tedious and time-consuming task. In this paper, we employ a capsule-based architecture to automate the task of macro- molecule identification, that we refer to as 3D-UCaps. In particular, the architecture is composed of three components: feature extractor, capsule encoder, and CNN decoder. The feature extractor converts voxel intensities of input sub-tomograms to activities of local features. The encoder is a 3D Capsule Network (CapsNet) that takes local features to generate a low-dimensional representation of the input. Then, a 3D CNN decoder reconstructs the sub-tomograms from the given representation by upsampling.
Results: We performed binary and multi-class localization and identification tasks on synthetic and experimental data. We observed that the 3D-UNet and the 3D-UCaps had an F1−score mostly above 60% and 70%, respectively, on the test data. In both network architectures, we observed degradation of at least 40% in the F1-score when identifying very small particles (PDB entry 3GL1) compared to a large particle (PDB entry 4D8Q). In the multi-class identification task of experimental data, 3D-UCaps had an F1-score of 91% on the test data in contrast to 64% of the 3D-UNet. The better F1-score of 3D-UCaps compared to 3D-UNet is obtained by a higher precision score. We speculate this to be due to the capsule network employed in the encoder. To study the effect of the CapsNet-based encoder architecture further, we performed an ablation study and perceived that the F1-score is boosted as network depth is increased which
is in contrast to the previously reported results for the 3D-UNet. To present a reproducible work, source code, trained models, data as well as visualization results are made publicly available.
Conclusion: Quantitative and qualitative results show that 3D-UCaps successfully perform various downstream tasks including identification and localization of macro- molecules and can at least compete with CNN architectures for this task. Given that the capsule layers extract both the existence probability and the orientation of the molecules, this architecture has the potential to lead to representations of the data that are better interpretable than those of 3D-UNet.
Understanding the Romanization Spreading on Historical Interregional Networks in Northern Tunisia
(2022)
Spreading processes are important drivers of change in social systems. To understand the mechanisms of spreading it is fundamental to have information about the underlying contact network and the dynamical parameters of the process.
However, in many real-wold examples, this information is not known and needs to be inferred from data. State-of-the-art spreading inference methods have mostly been applied to modern social systems, as they rely on availability of very detailed data. In this paper we study the inference challenges for historical spreading processes, for which only very fragmented information is available. To cope with this problem, we extend existing network models by formulating a model on a mesoscale with temporal spreading rate. Furthermore, we formulate the respective parameter inference problem for the extended model. We apply our approach to the romanization process of Northern Tunisia, a scarce dataset, and study properties of the inferred time-evolving interregional networks. As a result, we show that (1) optimal solutions consist of very different network structures and spreading rate functions; and that (2) these diverse solutions produce very similar spreading patterns. Finally, we discuss how inferred dominant interregional connections are related to available archaeological traces. Historical networks resulting from our approach can help understanding complex processes of cultural change in ancient times.
Understanding the Romanization Spreading on Historical Interregional Networks in Northern Tunisia
(2022)
Spreading processes are important drivers of change in social systems. To understand the mechanisms of spreading it is fundamental to have information about the underlying contact network and the dynamical parameters of the process. However, in many real-wold examples, this information is not known and needs to be inferred from data. State-of-the-art spreading inference methods have mostly been applied to modern social systems, as they rely on availability of very detailed data. In this paper we study the inference challenges for historical spreading processes, for which only very fragmented information is available. To
cope with this problem, we extend existing network models by formulating a model on a mesoscale with temporal spreading rate. Furthermore, we formulate the respective parameter inference problem for the extended model. We apply our approach to the romanization process of Northern Tunisia, a scarce dataset, and study properties of the inferred time-evolving interregional networks. As a result, we show that (1) optimal solutions consist of very different network structures and spreading rate functions; and that (2) these diverse solutions produce very similar spreading patterns. Finally, we discuss how inferred dominant interregional connections are related to available archaeological traces. Historical networks resulting from our approach can help understanding complex processes of cultural
change in ancient times.
In this paper, we explore the relationship patterns between Ancient Egyptian texts of the corpus ``Synodal decrees'', which are originating between 243 and 185 BCE, during the Ptolemaic period. Particularly, we are interested in analyzing the grammatical features of the different texts. Conventional data analysis methods such as correspondence Analysis are very useful to explore the patterns of statistical interdependence between categories of variables. However, it is based on a PCA-like dimension-reduction method and turned out to be unsuitable for our dataset due to the high dimensionality of our data representations. Additionally, the similarity between pairs of texts and pairs of grammatical features is observed through the distance between their representation, but the degree of association between a particular grammatical feature and a text is not. Here, we applied a qualitative Euclidean embedding method that provides a new Euclidean representation of the categories of variables. This new representation of the categories is constructed in such a way that all the patterns of statistical interdependence, similarity, and association, are seen through the Euclidean distance between them. Nevertheless, the PCA-like dimension-reduction method also performed poorly on our new representation. Therefore, we obtained a two-dimensional visualization using non-linear methods such UMAP or t-SNE. Although these dimension-reduction methods reduced the interpretability of interpoint distances, we were still able to identify important similarity patterns between the Synodal text as well as their association patterns with the grammatical features.