Identifying temporal and spatial patterns of variation from multi-modal data using MEFISTO

AbstractFactor analysis is among the most-widely used methods for dimensionality reduction in genome biology, with applications from personalized health to single-cell studies. Existing implementations of factor analysis assume independence of the observed samples, an assumption that fails in emerging spatio-temporal profiling studies. Here, we present MEFISTO, a flexible and versatile toolbox for modelling high-dimensional data when spatial or temporal dependencies between the samples are known. MEFISTO maintains the established benefits of factor analysis for multi-modal data, but enables performing spatio-temporally informed dimensionality reduction, interpolation and separation of smooth from non-smooth patterns of variation. Moreover, MEFISTO can integrate multiple related datasets by simultaneously identifying and aligning the underlying patterns of variation in a data-driven manner. We demonstrate MEFISTO through applications to an evolutionary atlas of mammalian organ development, where the model reveals conserved and evolutionary diverged developmental programs. In applications to a longitudinal microbiome study in infants, birth mode and diet were highlighted as major causes for heterogeneity in the temporally-resolved microbiome over the first years of life. Finally, we demonstrate that the proposed framework can also be applied to spatially resolved transcriptomics.

Download Full-text

Identifying temporal and spatial patterns of variation from multimodal data using MEFISTO

Nature Methods ◽

10.1038/s41592-021-01343-9 ◽

2022 ◽

Author(s):

Britta Velten ◽

Jana M. Braunger ◽

Ricard Argelaguet ◽

Damien Arnol ◽

Jakob Wirbel ◽

...

Keyword(s):

Factor Analysis ◽

Dimensionality Reduction ◽

Single Cell ◽

Cell Biology ◽

Multimodal Data ◽

Spatially Resolved ◽

Temporal And Spatial ◽

Spatio Temporal ◽

Personalized Health ◽

Analysis Models

AbstractFactor analysis is a widely used method for dimensionality reduction in genome biology, with applications from personalized health to single-cell biology. Existing factor analysis models assume independence of the observed samples, an assumption that fails in spatio-temporal profiling studies. Here we present MEFISTO, a flexible and versatile toolbox for modeling high-dimensional data when spatial or temporal dependencies between the samples are known. MEFISTO maintains the established benefits of factor analysis for multimodal data, but enables the performance of spatio-temporally informed dimensionality reduction, interpolation, and separation of smooth from non-smooth patterns of variation. Moreover, MEFISTO can integrate multiple related datasets by simultaneously identifying and aligning the underlying patterns of variation in a data-driven manner. To illustrate MEFISTO, we apply the model to different datasets with spatial or temporal resolution, including an evolutionary atlas of organ development, a longitudinal microbiome study, a single-cell multi-omics atlas of mouse gastrulation and spatially resolved transcriptomics.

Download Full-text

Physical-oriented and machine learning-based emission modeling in a diesel compression ignition engine: Dimensionality reduction and regression

International Journal of Engine Research ◽

10.1177/14680874211070736 ◽

2022 ◽

pp. 146808742110707

Author(s):

Aran Mohammad ◽

Reza Rezaei ◽

Christopher Hayduk ◽

Thaddaeus Delebinski ◽

Saeid Shahpouri ◽

...

Keyword(s):

Principal Component Analysis ◽

Support Vector Machine ◽

Factor Analysis ◽

Dimensionality Reduction ◽

Principal Component ◽

Component Analysis ◽

Data Driven ◽

Support Vector ◽

Emission Models ◽

Emission Modeling

The development of internal combustion engines is affected by the exhaust gas emissions legislation and the striving to increase performance. This demands for engine-out emission models that can be used for engine optimization for real driving emission controls. The prediction capability of physically and data-driven engine-out emission models is influenced by the system inputs, which are specified by the user and can lead to an improved accuracy with increasing number of inputs. Thereby the occurrence of irrelevant inputs becomes more probable, which have a low functional relation to the emissions and can lead to overfitting. Alternatively, data-driven methods can be used to detect irrelevant and redundant inputs. In this work, thermodynamic states are modeled based on 772 stationary measured test bench data from a commercial vehicle diesel engine. Afterward, 37 measured and modeled variables are led into a data-driven dimensionality reduction. For this purpose, approaches of supervised learning, such as lasso regression and linear support vector machine, and unsupervised learning methods like principal component analysis and factor analysis are applied to select and extract the relevant features. The selected and extracted features are used for regression by the support vector machine and the feedforward neural network to model the NOx, CO, HC, and soot emissions. This enables an evaluation of the modeling accuracy as a result of the dimensionality reduction. Using the methods in this work, the 37 variables are reduced to 25, 22, 11, and 16 inputs for NOx, CO, HC, and soot emission modeling while maintaining the accuracy. The features selected using the lasso algorithm provide more accurate learning of the regression models than the extracted features through principal component analysis and factor analysis. This results in test errors RMSETe for modeling NOx, CO, HC, and soot emissions 19.22 ppm, 6.46 ppm, 1.29 ppm, and 0.06 FSN, respectively.

Download Full-text

Structured Bayesian Gaussian process latent variable model: Applications to data-driven dimensionality reduction and high-dimensional inversion

Journal of Computational Physics ◽

10.1016/j.jcp.2018.12.037 ◽

2019 ◽

Vol 383 ◽

pp. 166-195 ◽

Cited By ~ 10

Author(s):

Steven Atkinson ◽

Nicholas Zabaras

Keyword(s):

Dimensionality Reduction ◽

Gaussian Process ◽

Latent Variable ◽

Latent Variable Model ◽

Data Driven ◽

High Dimensional ◽

Variable Model

Download Full-text

Factor analysis of temporal and spatial patterns of multicellular green freshwater algae

SIL Proceedings 1922-2010 ◽

10.1080/03680770.1977.11896875 ◽

1978 ◽

Vol 20 (4) ◽

pp. 2341-2346

Author(s):

H. Hillebrand

Keyword(s):

Factor Analysis ◽

Spatial Patterns ◽

Freshwater Algae ◽

Temporal And Spatial Patterns ◽

Temporal And Spatial

Download Full-text

Ultrastructural analysis of unique glycoconjugates in the rat olfactory system

Proceedings, annual meeting, Electron Microscopy Society of America ◽

10.1017/s0424820100124859 ◽

1992 ◽

Vol 50 (1) ◽

pp. 890-891

Author(s):

James E. Crandall ◽

Linda C. Hassinger ◽

Gerald A. Schwarting

Keyword(s):

Central Nervous System ◽

Nervous System ◽

Monoclonal Antibodies ◽

Olfactory System ◽

Olfactory Bulbs ◽

Temporal And Spatial Patterns ◽

Carbohydrate Epitopes ◽

Olfactory Systems ◽

Temporal And Spatial ◽

Cell Surface Glycoconjugates

Cell surface glycoconjugates are considered to play important roles in cell-cell interactions in the developing central nervous system. We have previously described a group of monoclonal antibodies that recognize defined carbohydrate epitopes and reveal unique temporal and spatial patterns of immunoreactivity in the developing main and accessory olfactory systems in rats. Antibody CC2 reacts with complex α-galactosyl and α-fucosyl glycoproteins and glycolipids. Antibody CC1 reacts with terminal N-acetyl galactosamine residues of globoside-like glycolipids. Antibody 1B2 reacts with β-galactosyl glycolipids and glycoproteins. Our light microscopic data suggest that these antigens may be located on the surfaces of axons of the vomeronasal and olfactory nerves as well as on some of their target neurons in the main and accessory olfactory bulbs.

Download Full-text

Hawk kite as potential bird scare device (the case of pigeons and grain processing factory)

Ukrainian Journal of Ecology ◽

10.15421/2018_350 ◽

2018 ◽

Vol 8 (2) ◽

pp. 334-336

Author(s):

A. V. Matsyura

Keyword(s):

Spatial Patterns ◽

Repellent Effect ◽

Preliminary Results ◽

Temporal And Spatial Patterns ◽

Grain Processing ◽

Feral Pigeons ◽

Temporal And Spatial ◽

Bird Behavior ◽

Hooded Crows

Here we presented the preliminary results of hawk kite usage against the feral pigeons in some grain processing factory. We studied the temporal and spatial patterns of repellent effect and bird behavior. We suggested the feral pigeons gradually increase the level of tolerance towards the hawk kite if no additional repellent measures were undertaken. Moreover, even initially the feral pigeons demonstrate higher tolerance towards the hawk kite compared to the Rooks or Hooded Crows.

Download Full-text

RHDSI: A Novel Dimensionality Reduction Based Algorithm on High Dimensional Feature Selection with Interactions

Information Sciences ◽

10.1016/j.ins.2021.06.096 ◽

2021 ◽

Author(s):

Rahi Jain ◽

Wei Xu

Keyword(s):

Feature Selection ◽

Dimensionality Reduction ◽

High Dimensional

Download Full-text

Evolution of Mouse T-box Genes by Tandem Duplication and Cluster Dispersion

Genetics ◽

10.1093/genetics/144.1.249 ◽

1996 ◽

Vol 144 (1) ◽

pp. 249-254 ◽

Cited By ~ 4

Author(s):

Sergei I Agulnik ◽

Nancy Garvey ◽

Sarah Hancock ◽

Ilya Ruvinsky ◽

Deborah L Chapman ◽

...

Keyword(s):

Tandem Duplication ◽

Crossing Over ◽

Dna Binding Domain ◽

Gene Products ◽

New Members ◽

Ancestral Gene ◽

T Box ◽

Temporal And Spatial Patterns ◽

Temporal And Spatial ◽

T Locus

Abstract The T-box genes comprise an ancient family of putative transcription factors conserved across species as divergent as Mus musculus and Caenorhabditis elegans. All T-box gene products are characterized by a novel 174-186amino acid DNA binding domain called the T-box that was first discovered in the polypeptide products of the mouse T locus and the Drosophila melanogaster optomotor-blind gene. Earlier studies allowed the identification of five mouse T-box genes, T, Tbx1-3, and Tbr1, that all map to different chromosomal locations and are expressed in unique temporal and spatial patterns during embryogenesis. Here, we report the discovery of three new members of the mouse T-box gene family, named Tbx4, Tbx5, and Tbx6. Two of these newly discovered genes, Tbx4 and Tbx5, were found to be tightly linked to previously identified T-box genes. Combined results from phylogenetic, linkage, and physical mapping studies provide a picture for the evolution of a T-box subfamily by unequal crossing over to form a two-gene cluster that was duplicated and dispersed to two chromosomal locations. This analysis suggests that Tbx4 and Tbx5 are cognate genes that diverged apart from a common ancestral gene during early vertebrate evolution.

Download Full-text