A clustering procedure for three-way RNA sequencing data using data transformations and matrix-variate Gaussian mixture models.

RNA sequencing of time-course experiments results in three-way count data where the dimensions are the genes, the time points and the biological units. Clustering RNA-seq data allows to extract groups of co-expressed genes over time. After standardisation, the normalised counts of individual genes across time points and biological units have similar properties as compositional data. We propose the following procedure to suitably cluster three-way RNA-seq data: (1) pre-process the RNA-seq data by calculating the normalised expression profiles, (2) transform the data using the additive log ratio transform to map the composition in the D-part Aitchison simplex to a D - 1 -dimensional Euclidean vector, (3) cluster the transformed RNA-seq data using matrix-variate Gaussian mixture models and (4) assess the quality of the overall cluster solution and of individual clusters based on cluster separation in the transformed space using density-based silhouette information and on compactness of the cluster in the original space using cluster maps as a suitable visualisation. The proposed procedure is illustrated on RNA-seq data from fission yeast and results are also compared to an analogous two-way approach after flattening out the biological units.

Sensors and chemometrics in downstream processing.

The biopharmaceutical industry is still running in batch mode, mostly because it is highly regulated. In the past, sensors were not readily available and in-process control was mainly executed offline. The most important product parameters are quantity, purity, and potency, in addition to adventitious agents and bioburden. New concepts using disposable single-use technologies and integrated bioprocessing for manufacturing will dominate the future of bioprocessing. To ensure the quality of pharmaceuticals, initiatives such as Process Analytical Technologies, Quality by Design, and Continuous Integrated Manufacturing have been established. The aim is that these initiatives, together with technology development, will pave the way for process automation and autonomous bioprocessing without any human intervention. Then, real-time release would be realized, leading to a highly predictive and robust biomanufacturing system. The steps toward such automated and autonomous bioprocessing are reviewed in the context of monitoring and control. It is possible to integrate real-time monitoring gradually, and it should be considered from a soft sensor perspective. This concept has already been successfully implemented in other industries and requires relatively simple model training and the use of established statistical tools, such as multivariate statistics or neural networks. This review describes a scenario for integrating soft sensors and predictive chemometrics into modern process control. This is exemplified by selective downstream processing steps, such as chromatography and membrane filtration, the most common unit operations for separation of biopharmaceuticals.

Prediction of the performance of pre-packed purification columns through machine learning.

Pre-packed columns have been increasingly used in process development and biomanufacturing thanks to their ease of use and consistency. Traditionally, packing quality is predicted through rate models, which require extensive calibration efforts through independent experiments to determine relevant mass transfer and kinetic rate constants. Here we propose machine learning as a complementary predictive tool for column performance. A machine learning algorithm, extreme gradient boosting, was applied to a large data set of packing quality (plate height and asymmetry) for pre-packed columns as a function of quantitative parameters (column length, column diameter, and particle size) and qualitative attributes (backbone and functional mode). The machine learning model offered excellent predictive capabilities for the plate height and the asymmetry (90 and 93%, respectively), with packing quality strongly influenced by backbone (∼70% relative importance) and functional mode (∼15% relative importance), well above all other quantitative column parameters. The results highlight the ability of machine learning to provide reliable predictions of column performance from simple, generic parameters, including strategic qualitative parameters such as backbone and functionality, usually excluded from quantitative considerations. Our results will guide further efforts in column optimization, for example, by focusing on improvements of backbone and functional mode to obtain optimized packings.

Technology transfer of a monitoring system to predict product concentration and purity of biopharmaceuticals in real-time during chromatographic separation.

Technological developments require the transfer to their location of application to make use of them. We describe the transfer of a real-time monitoring system for lab-scale preparative chromatography to two new sites where it will be used and developed further. Equivalent equipment was used. The capture of a biopharmaceutical model protein, human fibroblast growth factor 2 (FGF-2) was used to evaluate the system transfer. Predictive models for five quality attributes based on partial least squares regression were transferred. Six out of seven online sensors (UV/VIS, pH, conductivity, IR, RI, and MALS) showed comparable signals between the sites while one sensor (fluorescence) showed different signal profiles. A direct transfer of the models for real-time monitoring was not possible, mainly due to differences in sensor signals. Adaptation of the models was necessary. Then, among five prediction models, the prediction errors of the test run at the new sites were on average twice as high as at the training site (model-wise 0.9-5.7 times). Additionally, new prediction models for different products were trained at each new site. These allowed monitoring the critical quality attributes of two new biopharmaceutical products during their purification processes with mean relative deviations between 1% and 33%.

Identification of PMN-released mutagenic factors in a co-culture model for colitis-associated cancer.

Microsatellite instability (MSI) is present in ulcerative colitis (UC) and colitis-associated colorectal cancers (CAC). Certain factors released by polymorphonuclear cells (PMNs) may drive mucosal frameshift mutations resulting in MSI and cancer. Here, we applied a co-culture system with PMNs and colon epithelial cells to identify such culprit factors. Subjecting HCT116 + chr3 and human colonic epithelial cells (HCEC)-1CT MSI-reporter cell lines harboring mono-, di- or tetranucleotide DNA repeats linked to enhanced green fluorescent protein (EGFP) to activated PMNs induced frameshift mutations within all repeats, as quantified by flow cytometry. Activated PMNs released superoxide and hydrogen peroxide (H2O2), as measured by lucigenin-amplified chemiluminescence and fluorometry, respectively. Catalase, which scavenges H2O2, reduced such PMN-induced MSI. The NADPH-oxidase inhibitor apocynin, which blocks the oxidative burst in PMNs, similarly inhibited PMN-induced MSI. A bead-based multiplex assay revealed that PMNs release a wide range of cytokines such as interleukin (IL)-8, IL-6 and tumor necrosis factor-α (TNF-α). In vitro, these cytokines increased MSI in colon epithelial cells, and the Janus kinase (JAK) inhibitor tofacitinib abolished IL-6-induced or PMN-induced MSI. Intracellular reactive oxygen species (ROS) formation, as measured by 2',7'-dichlorofluorescein diacetate (DCFDA) assay, was induced upon cytokine treatment. DNA oxidation upon IL-6 was present, as detected by formamidopyrimidine glycosylase (FPG)-modified comet assay. In conclusion, activated PMNs induce frameshift mutations in colon epithelial cells resulting in MSI. Both oxidative burst with release of ROS and PMN-secreted cytokines, such as IL-8, IL-6 or TNF-α, contribute to MSI. ROS scavengers and/or specific inhibitors of cytokine signaling may delay or prevent cancer development in the setting of colitis.

Boosted structured additive regression for Escherichia coli fed-batch fermentation modeling.

The quality of biopharmaceuticals and patients' safety are of highest priority and there are tremendous efforts to replace empirical production process designs by knowledge-based approaches. Main challenge in this context is that real-time access to process variables related to product quality and quantity is severely limited. To date comprehensive on- and offline monitoring platforms are used to generate process data sets that allow for development of mechanistic and/or data driven models for real-time prediction of these important quantities. Ultimate goal is to implement model based feed-back control loops that facilitate online control of product quality. In this contribution, we explore structured additive regression (STAR) models in combination with boosting as a variable selection tool for modeling the cell dry mass, product concentration, and optical density on the basis of online available process variables and two-dimensional fluorescence spectroscopic data. STAR models are powerful extensions of linear models allowing for inclusion of smooth effects or interactions between predictors. Boosting constructs the final model in a stepwise manner and provides a variable importance measure via predictor selection frequencies. Our results show that the cell dry mass can be modeled with a relative error of about ±3%, the optical density with ±6%, the soluble protein with ±16%, and the insoluble product with an accuracy of ±12%. Biotechnol. Bioeng. 2017;114: 321-334. © 2016 Wiley Periodicals, Inc.

Mesalazine and thymoquinone attenuate intestinal tumour development in Msh2(loxP/loxP) Villin-Cre mice.

OBJECTIVE: Lynch syndrome is caused by germline mutations in DNA mismatch repair genes leading to microsatellite instability (MSI) and colorectal cancer. Mesalazine, commonly used for the treatment of UC, reduces MSI in vitro. Here, we tested natural compounds for such activity and applied mesalazine and thymoquinone in a Msh2(loxP/loxP) Villin-Cre mouse model for Lynch syndrome. DESIGN: Flow cytometry was used for quantitation of mutation rates at a CA13 microsatellite in human colon cancer (HCT116) cells that had been stably transfected with pIREShyg2-enhanced green fluorescent protein/CA13, a reporter for frameshift mutations. Mice were treated for 43 weeks with mesalazine, thymoquinone or control chow. Intestines were analysed for tumour incidence, tumour multiplicity and size. MSI testing was performed from microdissected normal intestinal or tumour tissue, compared with mouse tails and quantified by the number of mutations per marker (NMPM). RESULTS: Besides mesalazine, thymoquinone significantly improved replication fidelity at 1.25 and 2.5 µM in HCT116 cells. In Msh2(loxP/loxP) Villin-Cre mice, tumour incidence was reduced by mesalazine from 94% to 69% (p=0.04) and to 56% (p=0.003) by thymoquinone. The mean number of tumours was reduced from 3.1 to 1.4 by mesalazine (p=0.004) and to 1.1 by thymoquinone (p<0.001). Interestingly, MSI was reduced in normal intestinal tissue from 1.5 to 1.2 NMPM (p=0.006) and to 1.1 NMPM (p=0.01) by mesalazine and thymoquinone, respectively. Thymoquinone, but not mesalazine, reduced MSI in tumours. CONCLUSIONS: Mesalazine and thymoquinone reduce tumour incidence and multiplicity in Msh2(loxP/loxP) Villin-Cre mice by reduction of MSI independent of a functional mismatch repair system. Both substances are candidate compounds for chemoprevention in Lynch syndrome mutation carriers.

Explainable deep learning enhances robust and reliable real-time monitoring of a chromatographic protein A capture step.

The application of model-based real-time monitoring in biopharmaceutical production is a major step toward quality-by-design and the fundament for model predictive control. Data-driven models have proven to be a viable option to model bioprocesses. In the high stakes setting of biopharmaceutical manufacturing it is essential to ensure high model accuracy, robustness, and reliability. That is only possible when (i) the data used for modeling is of high quality and sufficient size, (ii) state-of-the-art modeling algorithms are employed, and (iii) the input-output mapping of the model has been characterized. In this study, we evaluate the accuracy of multiple data-driven models in predicting the monoclonal antibody (mAb) concentration, double stranded DNA concentration, host cell protein concentration, and high molecular weight impurity content during elution from a protein A chromatography capture step. The models achieved high-quality predictions with a normalized root mean squared error of <4% for the mAb concentration and of ≈10% for the other process variables. Furthermore, we demonstrate how permutation/occlusion-based methods can be used to gain an understanding of dependencies learned by one of the most complex data-driven models, convolutional neural network ensembles. We observed that the models generally exhibited dependencies on correlations that agreed with first principles knowledge, thereby bolstering confidence in model reliability. Finally, we present a workflow to assess the model behavior in case of systematic measurement errors that may result from sensor fouling or failure. This study represents a major step toward improved viability of data-driven models in biopharmaceutical manufacturing.

Modelling time course gene expression data with finite mixtures of linear additive models.

UNLABELLED: A model class of finite mixtures of linear additive models is presented. The component-specific parameters in the regression models are estimated using regularized likelihood methods. The advantages of the regularization are that (i) the pre-specified maximum degrees of freedom for the splines is less crucial than for unregularized estimation and that (ii) for each component individually a suitable degree of freedom is selected in an automatic way. The performance is evaluated in a simulation study with artificial data as well as on a yeast cell cycle dataset of gene expression levels over time. AVAILABILITY: The latest release version of the R package flexmix is available from CRAN (http://cran.r-project.org/).

Comparative transcription profiling and in-depth characterization of plasmid-based and plasmid-free Escherichia coli expression systems under production conditions.

Plasmid-based Escherichia coli BL21(DE3) expression systems are extensively used for the production of recombinant proteins. However, the combination of a high gene dosage with strong promoters exerts extremely stressful conditions on producing cells, resulting in a multitude of protective reactions and malfunctions in the host cell with a strong impact on yield and quality of the product. Here, we provide in-depth characterization of plasmid-based perturbations in recombinant protein production. A plasmid-free T7 system with a single copy of the gene of interest (GOI) integrated into the genome was used as a reference. Transcriptomics in combination with a variety of process analytics were used to characterize and compare a plasmid-free T7-based expression system to a conventional pET-plasmid-based expression system, with both expressing human superoxide dismutase in fed-batch cultivations. The plasmid-free system showed a moderate stress response on the transcriptional level, with only minor effects on cell growth. In contrast to this finding, comprehensive changes on the transcriptome level were observed in the plasmid-based expression system and cell growth was heavily impaired by recombinant gene expression. Additionally, we found that the T7 terminator is not a sufficient termination signal. Overall, this work reveals that the major metabolic burden in plasmid-based systems is caused at the level of transcription as a result of overtranscription of the multicopy product gene and transcriptional read-through of T7 RNA polymerase. We therefore conclude that the presence of high levels of extrinsic mRNAs, competing for the limited number of ribosomes, leads to the significantly reduced translation of intrinsic mRNAs.

Spoilage markers for freshwater fish: A comprehensive workflow for non-targeted analysis of VOCs using DHS-GC-HRMS.

Changes of volatile organic compounds (VOCs) patterns during 6 days of storage at +4 °C were investigated in different freshwater fish species, namely carp and trout, using dynamic headspace gas chromatography time-of-flight mass spectrometry (DHS-GC-TOFMS). DHS parameters were systematically optimized to establish optimum extraction and pre-concentration of VOCs. Moreover, different sample preparation methods were tested: mincing with a manual meat grinder, as well as mincing plus homogenization with a handheld homogenizer both without and with water addition. The addition of water during sample preparation led to pronounced changes of the volatile profiles, depending on the molecular structure and lipophilicity of the analytes, resulting in losses of up to 98 % of more lipophilic compounds (logP > 3). The optimized method was applied to trout and carp. Trout samples of different storage days were compared using univariate (Mann-Whitney U test, fold change calculation) and multivariate (OPLS-DA) statistics. 37 potential spoilage markers were selected; for 11 compounds identity could be confirmed via measurement of authentic standards and 10 compounds were identified by library spectrum match. 22 compounds were also found to be statistically significant spoilage markers in carp. Merging results of the different statistical approaches, the list of 37 compounds could be narrowed down to the 14 most suitable for trout spoilage assessment. This study comprises a systematic evaluation of the capabilities of DHS-GC coupled to high-resolution (HR) MS for studying spoilage in different freshwater fish species, including a comprehensive data evaluation workflow.

The nucleotide composition of microsatellites impacts both replication fidelity and mismatch repair in human colorectal cells.

Microsatellite instability is a key mechanism of colon carcinogenesis. We have previously studied mutations within a (CA)13 microsatellite using an enhanced green fluorescent protein (EGFP)-based reporter assay that allows the distinction of replication errors and mismatch repair (MMR) activity. Here we utilize this assay to compare mutations of mono- and dinucleotide repeats in human colorectal cells. HCT116 and HCT116+chr3 cells were stably transfected with EGFP-based plasmids harboring A10, G10, G16, (CA)13 and (CA)26 repeats. EGFP-positive mutant fractions were quantitated by flow cytometry, mutation rates were calculated and the mutant spectrum was analyzed by cycle sequencing. EGFP fluorescence pattern changed with the microsatellite's nucleotide sequence and cell type and clonal variations were observed in mononucleotide repeats. Replication errors (as calculated in HCT116) at A10 repeats were 5-10-fold higher than in G10, G16 were 30-fold higher than G10 and (CA)26 were 10-fold higher than (CA)13. The mutation rates in hMLH1-proficient HCT116+chr3 were 30-230-fold lower than in HCT116. MMR was more efficient in G16 than in A10 clones leading to a higher stability of poly-G tracts. Mutation spectra revealed predominantly 1-unit deletions in A10, (CA)13 and G10 and 2-unit deletions or 1-unit insertion in (CA)26. These findings indicate that both replication fidelity and MMR are affected by the microsatellite's nucleotide composition.

Mixtures of regression models for time course gene expression data: evaluation of initialization and random effects.

SUMMARY: Finite mixture models are routinely applied to time course microarray data. Due to the complexity and size of this type of data, the choice of good starting values plays an important role. So far initialization strategies have only been investigated for data from a mixture of multivariate normal distributions. In this work several initialization procedures are evaluated for mixtures of regression models with and without random effects in an extensive simulation study on different artificial datasets. Finally, these procedures are also applied to a real dataset from Escherichia coli. AVAILABILITY: The latest release versions of R packages flexmix, gcExplorer and kernlab are always available from CRAN (http://cran.r-project.org/). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

gcExplorer: interactive exploration of gene clusters.

Cluster analysis plays an important role in the analysis of gene expression data since the early beginning of microarray studies and is routinely used to find groups of genes with common expression pattern. In order to make cluster analysis helpful for users, visualization of cluster solutions is of utmost importance. Here, we present the new R package gcExplorer for the interactive exploration of gene clusters. gcExplorer facilitates the interpretation of cluster results and allows to investigate extensive information about clusters.

Proteomics analysis of host cell proteins after immobilized metal affinity chromatography: Influence of ligand and metal ions.

Different degrees of protein purity have been observed in immobilized metal affinity chromatography ranging from extremely high purity to moderate and low purity. It has been hypothesized that the host cell protein composition and the metal ligands are factors governing the purity of a protein obtained after immobilized metal affinity chromatography (IMAC). Ni nitrilotriacetic acid (NTA) has become the first choice for facile His-tagged protein purification, but alternative ligands such as iminodiacetic acid (IDA) with other immobilized metal ions such as Zn, Cu and Co are valuable options when the expected purity or binding capacity is not reached. Especially Cu and Zn are very attractive, due to their reduced environmental and safety concerns compared to Ni. Co and Zn are more selective than Ni and Cu. This increased selectivity comes at the cost of weaker binding. In this work, the influence of ligand choice on protein purity after IMAC was evaluated by several methods, including peptide mapping. His-tagged GFP was used as model protein. We found that host cell protein (HCP) content varies drastically between ligands, as IDA eluates generally showing higher HCP concentrations than NTA. The relative content of the key amino acids His, Cys and Trp in the sequence of the co-eluted protein does not suffice to explain co-eluting propensity. The co-elution of HCPs is mostly influenced by metal binding clusters on the protein surface and not by total content or surface concentration of metal interacting amino acids. Prediction of co-elution is not dependent on these clusters alone, due to protein-protein interactions, indicted by a relative low metal binding cluster score but high co-elution propensity and in a lot of cases these proteins are often part of complex such as ribosome and chaperones. The different co-eluting proteins were presented by a heatmap with a dendrogram. Ward's linkage method was used to calculate the distance between groups of co-eluting proteins. Clustering of co-eluting HCPs was observed according to ligand and by metal ions, with Zn and Co forming one cluster and Ni and Cu another. The co-elution of host cell proteins can be explained by clusters of metal interacting amino acids on the protein surface and by protein-protein interactions. While Ni NTA still appears to be highly advantageous, it might not be the cure-all for all applications.

Exploratory and inferential analysis of gene cluster neighborhood graphs.

BACKGROUND: Many different cluster methods are frequently used in gene expression data analysis to find groups of co-expressed genes. However, cluster algorithms with the ability to visualize the resulting clusters are usually preferred. The visualization of gene clusters gives practitioners an understanding of the cluster structure of their data and makes it easier to interpret the cluster results. RESULTS: In this paper recent extensions of R package gcExplorer are presented. gcExplorer is an interactive visualization toolbox for the investigation of the overall cluster structure as well as single clusters. The different visualization options including arbitrary node and panel functions are described in detail. Finally the toolbox can be used to investigate the quality of a given clustering graphically as well as theoretically by testing the association between a partition and a functional group under study. CONCLUSION: It is shown that gcExplorer is a very helpful tool for a general exploration of microarray experiments. The identification of potentially interesting gene candidates or functional groups is substantially accelerated and eased. Inferential analysis on a cluster solution is used to judge its ability to provide insight into the underlying mechanistic biology of the experiment.

Interactive visualization of clusters in microarray data: an efficient tool for improved metabolic analysis of E. coli.

BACKGROUND: Interpretation of comprehensive DNA microarray data sets is a challenging task for biologists and process engineers where scientific assistance of statistics and bioinformatics is essential. Interdisciplinary cooperation and concerted development of software-tools for simplified and accelerated data analysis and interpretation is the key to overcome the bottleneck in data-analysis workflows. This approach is exemplified by gcExplorer an interactive visualization toolbox based on cluster analysis. Clustering is an important tool in gene expression data analysis to find groups of co-expressed genes which can finally suggest functional pathways and interactions between genes. The visualization of gene clusters gives practitioners an understanding of the cluster structure of their data and makes it easier to interpret the cluster results. RESULTS: In this study the interactive visualization toolbox gcExplorer is applied to the interpretation of E. coli microarray data. The data sets derive from two fedbatch experiments conducted in order to investigate the impact of different induction strategies on the host metabolism and product yield. The software enables direct graphical comparison of these two experiments. The identification of potentially interesting gene candidates or functional groups is substantially accelerated and eased. CONCLUSION: It was shown that gcExplorer is a very helpful tool to gain a general overview of microarray experiments. Interesting gene expression patterns can easily be found, compared among different experiments and combined with information about gene function from publicly available databases.

Prediction of the Quantity and Purity of an Antibody Capture Process in Real Time.

Regulatory recommendations for quality by design instead of quality by testing raise increasing interest in new sensor technologies. An online monitoring system for downstream processes is developed, which is based on an array of online detectors. Besides standard detectors (UV, pH, and conductivity), our chromatographic workstation is equipped with a fluorescence and a mid-infrared spectrometer, a light scattering, and a refractive index detector. The combination of these sensors enables the prediction of specific protein concentration and various purity attributes, such as high molecular weight impurities, DNA and host cell protein content during the elution phase of a chromatographic antibody capture process. Prediction models solely based on online signals are set up providing real-time predictions. No mechanistic models or information about the chromatographic runs is used. These predictions allow online pooling decisions replacing time- and labor-intensive laboratory measurements. Different process variations, such as changes in the column load or elution buffer, are introduced to test the predictive power of the models. Extrapolation of the models worked well when the column load is changed, whereas model adjustment is necessary when the elution conditions are changed considerably.

Additional support for linkage of schizophrenia and bipolar disorder to chromosome 3q29.

After publishing a genome scan and follow-up fine mapping, suggesting schizophrenia and bipolar disorder linkage to chromosome 3q29, we now genotyped 11 additional SNPs (single nucleotide polymorphisms), in order to narrow down a potential candidate region. Linkage was performed using the GENEHUNTER program version 2.1r3. A NPL score Z(all) of 3.891 (p=0.000156) was observed with SNP rs225. In short, we found significant linkage scores most telomeric on chromosome 3q29, spanning 3.46 Mbp (7 SNPs).

Trend analysis of performance parameters of pre-packed columns for protein chromatography over a time span of ten years.

Pre-packed small scale chromatography columns are increasingly used for process development, for determination of design space in bioprocess development, and for post-licence process verifications. The packing quality of 30,000 pre-packed columns delivered to customers over a period 10 years has been analyzed by advanced statistical tools. First, the data were extracted and checked for inconsistencies, and then were tabulated and made ready for statistical processing using the programming language Perl (https://www.perl.org/) and the statistical computing environment R (https://www.r-project.org/). Reduced HETP and asymmetry were plotted over time to obtain a trend of packing quality over 10 years. The obtained data were used as a visualized coefficient of variation analysis (VCVA), a process that has often been applied in other industries such as semiconductor manufacturing. A typical fluctuation of reduced HETP was seen. A Tsunami effect in manufacturing, the effect of propagation of manufacturing deviations leading to out-of-specification products, was not observed with these pre-packed columns. Principal component analysis (PCA) showed that all packing materials cluster. Our data analysis showed that the current commercially available chromatography media used for biopharmaceutical manufacturing can be reproducibly and uniformly packed in polymer-based chromatography columns, which are designed for ready-to-use purposes. Although the number of packed columns has quadrupled over one decade the packing quality has remained stable.