Assays

Usage of fine-tuned BioBERT for identification of chemical entities

Using semantic search in MesH and PubChem databases for entity linking

Harvest optical densites and methods of transcriptom, proteom and metabolom samples for all tested substrat conditions.

All flux balance analyses of the stoichiometric model

All files for metabolic modeling using the Metano toolbox, including the scenarios and flux balance analyses.

For scRNA-Seq, iSABs were dissociated using the Primary Cardiomyocyte Isolation Kit (Thermo Fisher Scientific) before library preparation was performed using the 10xGenomics system with subsequent sequencing on the HighSeq4000 (Illumina). The mouse-SAN scRNA-Seq protocol is described in Goodyer et al. Preprocessing of raw sequencing data from iSABs relied on tools of the Cell Ranger Software (v.6.1.0) as was the procedure in Goodyer et al. Downstream analyses were conducted similar for both ...

Application of the LoRAS oversampling approach on single-cell/single-nuclei data to annotate/identify specific cell populations in new data based on previously, manually curated data.

Here, we conduct a proof of principle by comparing a 2D and 3D fluorescent image analysis based approach on unlabeled cardiomyocytes. Based on the CellProfiler software, we extracted high-dimensional features of individual cells and nuclei, which are subsequently down-sampled and clustered. These clusters are furthermore benchmarked via different machine learning classifiers (e.g., AdaBoost, Gradient Boosting, Random Forest) as the ground truth for our proposed approach.

Simulation of OE mutants targetting enzymes in the model, combined with metabolite concentrations and enzyme fold change of from the 40 samples. For each second mutant the enzyme concentrations in case of OE and KO mutants in updated and the metabolite concentrations of the second sample are loaded in the model. Using this approach the model approximately predicts combinatorial effects of OE mutations with other mutations, perturbations and time series concentrations.

The associated zip files contains all input files and a Jupyter notebook to rerun sampled simmulations, combined simmulations, parameter scan for the model with addition of an oxygin inhibiton of LDH, local- and global-sensitivity analysis and plot simmulation output in various formats. In addition the zip file contains the py36.yaml file that can be used to recreate the model simmulation environment using Anaconda making all simmulations completely reproducable. All information on how to use ...

Training of the model, parameter estimation using Evolutionary Programming using metabolomics, proteomics and some flux data.

Validation by simulating independent OE, KO mutant and perturbation samples, using sampling of the gausian distribution based on the mean and SD of measurements per sample. A 1000 samples of the gausian distribution of the mean and SD was performed per sample to show error in the measurements and how it propegates in predicted metabolite concentration in SS

Protein copy number at 6h, 12h, 24h, 48h, 72h, 96h, average values and SD for the measurements

Metabolomics time series measurements for internal metabolites for 6h, 24h and 48h for multiple experiments. Largely based on MAss spectrometry, bioluminescence kits to measure NAD, NADH at 24h, other time points are infered from relative measurements times the absolute measurements at 24h.

Measurements of external metabolites based on growth curve data. Flux estimates for uptake of external metabolites such as glucose and production rates for external metabolites lactate and acetate

Metabolic control analysis: Local control coefficients for 40 independent samples based on 100x sampling from the measurement distribution Global control analysis based on 100.000 Latin Hypercube sampling from the parameter search range (0.01-100 for Km values and 0.001-1000 for Vmax values)