Journal: bioRxiv

Article Title: Multi-tissue transcriptomic aging atlas reveals predictive aging biomarkers in the killifish

doi: 10.1101/2025.01.28.635350

Figure Lengend Snippet: (a) Workflow of BayesAge 2.0, a Bayesian and locally weighted scatterplot smoothing (LOWESS) regression model behind the aging clocks. To train a tissue clock, Leave One Sample Out Cross-Validation (LOSO-CV) was used to generate testing-training splits of the data. In each iteration of LOSO-CV, one sample was used as a test set, while the rest of the tissue samples were used for training. This was performed k times, where k is the number of tissue samples available. Each time LOSO-CV was performed, a set of top age-associated genes (the highest absolute Spearman’s rank correlation values) was selected for the feature set. Then, the probability that the sample in the test set was a given age was calculated from the probability of the observed expression value for each selected gene in the sample at that age, assuming a Poisson distribution. The product of each gene-wise probability was computed to determine the age probability. The result was an age-probability distribution from which the age prediction was the highest probability age in this distribution. (b) Bar plots of the performance metrics for the BayesAge sex-combined tissue clocks, using the coefficient of determination (R 2 ) for the relationship between chronological and predicted age and the mean absolute error (MAE). (c) Scatterplot of gut clock chronological age vs. the ‘transcriptomic age’ (tAge) for measuring the prediction accuracy of the highest performing gut sex-combined tissue clock. The ‘optimal’ BayesAge clock is defined as the model with the most concordance between chronological and predicted age among all the gene number tested. Bottom, the gene frequency scatterplots of the top 10 overall age-correlated genes trained on the sex-combined gut samples are shown. The pink line is the locally estimated scatterplot smoothing (LOESS) regression fit across time. (d) Bar plots of R 2 and MAE values for select clocks trained on sex-combined data (left, ‘S-C’), female data (middle, ‘F’), and male data (right, ‘M’). Selected tissues include highly transcriptionally sex-dimorphic tissues (gonad, kidney, liver), moderately transcriptionally sex-dimorphic tissues (gut, skin), and one weakly sex-dimorphic tissue (brain). (e) Accuracy of tAge predictions for the optimal sex-combined (left), male-only (middle), and female-only liver clocks (right). (f) Predicted ages for liver samples from male and female killifish fed on ad libitum (AL) or dietary restricted (DR) diets using sex-dimorphic liver clocks (data from a published dataset ). Age prediction was performed using three different modeling strategies, BayesAge 2.0 (left), Elastic Net regression (middle), and Principal Component regression (right). Each dot in each box plot represents the predicted tAge for the liver transcriptome of an individual fish (4 fish per condition) and the gene set size or number of principal components used for age prediction is listed. For each model, Mann-Whitney test was used to test the significance of difference between the AL and DR conditions.

Article Snippet: This method utilizes a Bayesian framework to estimate the most likely transcriptomic age of a sample (‘tAge’) and employs locally weighted scatterplot smoothing (LOWESS) regression to model the nonlinear dynamics of gene expression, enabling age prediction between 47 to 163 days of age at day-level resolution.

Techniques: Biomarker Discovery, Expressing, MANN-WHITNEY

Journal: bioRxiv

Article Title: Multi-tissue transcriptomic aging atlas reveals predictive aging biomarkers in the killifish

doi: 10.1101/2025.01.28.635350

Figure Lengend Snippet: (a) Scatterplot of the tissue transcriptomic age (tAge) vs. chronological age for measuring the prediction accuracy of the optimal brain sex-combined tissue clock, which is the model that corresponds to the most concordance between chronological and predicted age among all the gene number tested. The coefficient of determination (R 2 ) between chronological and predicted age, as well as the mean absolute error (MAE), is listed in graphs. (b) The gene frequency scatterplots of the top 10 overall age-correlated genes trained on the sex-combined brain samples are shown. The black line is the locally weighted scatterplot smoothing (LOWESS) regression fit across time. (c, d) The scatterplots of tAge vs. chronological age (c) and gene frequency (d) were generated as in panels a and b, but for the testis.

Article Snippet: This method utilizes a Bayesian framework to estimate the most likely transcriptomic age of a sample (‘tAge’) and employs locally weighted scatterplot smoothing (LOWESS) regression to model the nonlinear dynamics of gene expression, enabling age prediction between 47 to 163 days of age at day-level resolution.

Techniques: Generated

Journal: Biotechnology and Bioengineering

Article Title: Segmented linear modeling of CHO fed‐batch culture and its application to large scale production

doi: 10.1002/bit.26214

Figure Lengend Snippet: Developed methodology to identify and characterize metabolic phases. Experimental data are first cleaned using the methodology presented in Figure 2 and additionally by removing data with a viability below 50% or a depletion of metabolites during a measurement interval. The number of metabolic phases during the cell culture process are determined by differentiating the smoothed (LOWESS) reaction rates of all metabolites with respect to the growth rate (dR/dµ). Recursive partitioning is then applied on those derivatives to get a vector of possible metabolic phase breakpoints. Hierarchical clustering is then applied on this vector of possible breakpoints to define the number of final metabolic phases (clusters). Knowing the number of metabolic phases, the segmented regression can then be calibrated on the calibration dataset for each metabolite and validated on the cross validation dataset of the 2 L bioreactor and also of the 2000 L bioreactor.

Article Snippet: As the derivative can amplify possible biological and analytical errors, the specific production rates were, preliminarily to deriving, smoothed as a function of the specific growth rate with the linear Locally Weighted Scatterplot Smoother (LOWESS) method (Cleveland, ) by using SAS software JMP 11 ©.

Techniques: Cell Culture, Plasmid Preparation, Biomarker Discovery