Controlling false discoveries in high-dimensional situations: boosting with stability selection

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-based variable selection for component-wise gradient boosting

Model-based component-wise gradient boosting is a popular tool for data-driven variable selection. In order to improve its prediction and selection qualities even further, several modifications of the original algorithm have been developed, that mainly focus on different stopping criteria, leaving the actual variable selection mechanism untouched. We investigate different prediction-based mechanisms for the variable selection step in model-based component-wise gradient boosting. These approaches include Akaikes Information Criterion (AIC) as well as a selection rule relying on the component-wise test error computed via cross-validation. We implemented the AIC and cross-validation routines for Generalized Linear Models and evaluated them regarding their variable selection properties and predictive performance. An extensive simulation study revealed improved selection properties whereas the prediction error could be lowered in a real world application with age-standardized COVID-19 incidence rates.

Post-translational modifications linked to preclinical Alzheimer's disease-related pathological and cognitive changes

INTRODUCTION

In this study, we leverage proteomic techniques to identify communities of proteins underlying Alzheimer's disease (AD) risk among clinically unimpaired (CU) older adults.

METHODS

We constructed a protein co-expression network using 3869 cerebrospinal fluid (CSF) proteins quantified by SomaLogic, Inc., in a cohort of participants along the AD clinical spectrum. We then replicated this network in an independent cohort of CU older adults and related these modules to clinically-relevant outcomes.

RESULTS

We discovered modules enriched for phosphorylation and ubiquitination that were associated with abnormal amyloid status, as well as p-tau181 (M4: β = 2.44, p < 0.001, M7: β = 2.57, p < 0.001) and executive function performance (M4: β = -2.00, p = 0.005, M7: β = -2.39, p < 0.001).

DISCUSSION

In leveraging CSF proteomic data from individuals spanning the clinical spectrum of AD, we highlight the importance of post-translational modifications for early cognitive and pathological changes.

Prognostic Factors of Perceived Disability and Perceived Recovery After Whiplash: A Longitudinal, Prospective Study With One-year Follow-up

OBJECTIVES

The understanding of the role that cognitive and emotional factors play in how an individual recovers from a whiplash injury is important. Hence, we sought to evaluate whether pain-related cognitions (self-efficacy beliefs, expectation of recovery, pain catastrophizing, optimism, and pessimism) and emotions (kinesiophobia) are longitudinally associated with the transition to chronic whiplash-associated disorders in terms of perceived disability and perceived recovery at 6 and 12 months.

METHODS

One hundred sixty-one participants with acute or subacute whiplash-associated disorder were included. The predictors were: self-efficacy beliefs, expectation of recovery, pain catastrophizing, optimism, pessimism, pain intensity, and kinesiophobia. The 2 outcomes were the dichotomized scores of perceived disability and recovery expectations at 6 and 12 months. Stepwise regression with bootstrap resampling was performed to identify the predictors most strongly associated with the outcomes and the stability of such selection.

RESULTS

Baseline perceived disability, pain catastrophizing, and expectation of recovery were the most likely to be statistically significant, with an overage frequency of 87.2%, 84.0%, and 84.0%, respectively.

CONCLUSION

Individuals with higher expectations of recovery and lower levels of pain catastrophizing and perceived disability at baseline have higher perceived recovery and perceived disability at 6 and 12 months. These results have important clinical implications as both factors are modifiable through health education approaches.

A boosting method to select the random effects in linear mixed models

This paper proposes a novel likelihood-based boosting method for the selection of the random effects in linear mixed models. The nonconvexity of the objective function to minimize, which is the negative profile log-likelihood, requires the adoption of new solutions. In this respect, our optimization approach also employs the directions of negative curvature besides the usual Newton directions. A simulation study and a real-data application show the good performance of the proposal.

Predicting tumour content of liquid biopsies from cell-free DNA

BACKGROUND

Liquid biopsy is a minimally-invasive method of sampling bodily fluids, capable of revealing evidence of cancer. The distribution of cell-free DNA (cfDNA) fragment lengths has been shown to differ between healthy subjects and cancer patients, whereby the distributional shift correlates with the sample's tumour content. These fragmentomic data have not yet been utilised to directly quantify the proportion of tumour-derived cfDNA in a liquid biopsy.

RESULTS

We used statistical learning to predict tumour content from Fourier and wavelet transforms of cfDNA length distributions in samples from 118 cancer patients. The model was validated on an independent dilution series of patient plasma.

CONCLUSIONS

This proof of concept suggests that our fragmentomic methodology could be useful for predicting tumour content in liquid biopsies.

Automatic Variable Selection Algorithms in Prognostic Factor Research in Neck Pain

This study aims to compare the variable selection strategies of different machine learning (ML) and statistical algorithms in the prognosis of neck pain (NP) recovery. A total of 3001 participants with NP were included. Three dichotomous outcomes of an improvement in NP, arm pain (AP), and disability at 3 months follow-up were used. Twenty-five variables (twenty-eight parameters) were included as predictors. There were more parameters than variables, as some categorical variables had >2 levels. Eight modelling techniques were compared: stepwise regression based on unadjusted p values (stepP), on adjusted p values (stepPAdj), on Akaike information criterion (stepAIC), best subset regression (BestSubset) least absolute shrinkage and selection operator [LASSO], Minimax concave penalty (MCP), model-based boosting (mboost), and multivariate adaptive regression splines (MuARS). The algorithm that selected the fewest predictors was stepPAdj (number of predictors, p = 4 to 8). MuARS was the algorithm with the second fewest predictors selected (p = 9 to 14). The predictor selected by all algorithms with the largest coefficient magnitude was "having undergone a neuroreflexotherapy intervention" for NP (β = from 1.987 to 2.296) and AP (β = from 2.639 to 3.554), and "Imaging findings: spinal stenosis" (β = from -1.331 to -1.763) for disability. Stepwise regression based on adjusted p-values resulted in the sparsest models, which enhanced clinical interpretability. MuARS appears to provide the optimal balance between model sparsity whilst retaining high predictive performance across outcomes. Different algorithms produced similar performances but resulted in a different number of variables selected. Rather than relying on any single algorithm, confidence in the variable selection may be increased by using multiple algorithms.

Plasma proteomic signatures of a direct measure of insulin sensitivity in two population cohorts

AIMS/HYPOTHESIS

The euglycaemic-hyperinsulinaemic clamp (EIC) is the reference standard for the measurement of whole-body insulin sensitivity but is laborious and expensive to perform. We aimed to assess the incremental value of high-throughput plasma proteomic profiling in developing signatures correlating with the M value derived from the EIC.

METHODS

We measured 828 proteins in the fasting plasma of 966 participants from the Relationship between Insulin Sensitivity and Cardiovascular disease (RISC) study and 745 participants from the Uppsala Longitudinal Study of Adult Men (ULSAM) using a high-throughput proximity extension assay. We used the least absolute shrinkage and selection operator (LASSO) approach using clinical variables and protein measures as features. Models were tested within and across cohorts. Our primary model performance metric was the proportion of the M value variance explained (R2).

RESULTS

A standard LASSO model incorporating 53 proteins in addition to routinely available clinical variables increased the M value R2 from 0.237 (95% CI 0.178, 0.303) to 0.456 (0.372, 0.536) in RISC. A similar pattern was observed in ULSAM, in which the M value R2 increased from 0.443 (0.360, 0.530) to 0.632 (0.569, 0.698) with the addition of 61 proteins. Models trained in one cohort and tested in the other also demonstrated significant improvements in R2 despite differences in baseline cohort characteristics and clamp methodology (RISC to ULSAM: 0.491 [0.433, 0.539] for 51 proteins; ULSAM to RISC: 0.369 [0.331, 0.416] for 67 proteins). A randomised LASSO and stability selection algorithm selected only two proteins per cohort (three unique proteins), which improved R2 but to a lesser degree than in standard LASSO models: 0.352 (0.266, 0.439) in RISC and 0.495 (0.404, 0.585) in ULSAM. Reductions in improvements of R2 with randomised LASSO and stability selection were less marked in cross-cohort analyses (RISC to ULSAM R2 0.444 [0.391, 0.497]; ULSAM to RISC R2 0.348 [0.300, 0.396]). Models of proteins alone were as effective as models that included both clinical variables and proteins using either standard or randomised LASSO. The single most consistently selected protein across all analyses and models was IGF-binding protein 2.

CONCLUSIONS/INTERPRETATION

A plasma proteomic signature identified using a standard LASSO approach improves the cross-sectional estimation of the M value over routine clinical variables. However, a small subset of these proteins identified using a stability selection algorithm affords much of this improvement, especially when considering cross-cohort analyses. Our approach provides opportunities to improve the identification of insulin-resistant individuals at risk of insulin resistance-related adverse health consequences.

Analysis of serum proteomics data identifies a quantitative association between beta-defensin 2 at baseline and clinical response to IL-17 blockade in psoriatic arthritis

OBJECTIVES

Despite several effective targeted therapies, biomarkers that predict whether a patient with psoriatic arthritis (PsA) will respond to a particular treatment are currently lacking.

METHODS

We analysed proteomics data from serum samples of nearly 2000 patients with PsA in placebo-controlled phase-III clinical trials of the interleukin-17 inhibitor secukinumab. To discover predictive biomarkers of clinical response, we used statistical learning with controlled feature selection. The top candidate was validated using an ELISA and was separately assessed in a trial of almost 800 patients with PsA treated with secukinumab or the tumour necrosis factor inhibitor adalimumab.

RESULTS

Serum levels of beta-defensin 2 (BD-2) at baseline were found to be robustly associated with subsequent clinical response (eg, American College of Rheumatology definition of 20%, 50% and 70% improvement) to secukinumab, but not to placebo. This finding was validated in two independent clinical studies not used for discovery. Although BD-2 is known to be associated with psoriasis severity, the predictivity of BD-2 was independent of baseline Psoriasis Area and Severity Index. The association between BD-2 and response to secukinumab was observed as early as 4 weeks and maintained up to 52 weeks. BD-2 was also found to predict response to treatment with adalimumab. Unlike in PsA, BD-2 was not predictive of response to secukinumab in rheumatoid arthritis.

CONCLUSIONS

In PsA, BD-2 at baseline is quantitatively associated with clinical response to secukinumab. Patients with high levels of BD-2 at baseline reach and sustain higher rates of clinical response after treatment with secukinumab.

Amplification of annual and diurnal cycles of alpine lightning

The response of lightning to a changing climate is not fully understood. Historic trends of proxies known for fostering convective environments suggest an increase of lightning over large parts of Europe. Since lightning results from the interaction of processes on many scales, as many of these processes as possible must be considered for a comprehensive answer. Recent achievements of decade-long seamless lightning measurements and hourly reanalyses of atmospheric conditions including cloud micro-physics combined with flexible regression techniques have made a reliable reconstruction of cloud-to-ground lightning down to its seasonally varying diurnal cycle feasible. The European Eastern Alps and their surroundings are chosen as reconstruction region since this domain includes a large variety of land-cover, topographical and atmospheric circulation conditions. The most intense changes over the four decades from 1980 to 2019 occurred over the high Alps where lightning activity doubled in the 2010 s compared to the 1980 s. There, the lightning season reaches a higher maximum and starts one month earlier. Diurnally, the peak is up to 50% stronger with more lightning strikes in the afternoon and evening hours. Signals along the southern and northern alpine rim are similar but weaker whereas the flatlands surrounding the Alps have no significant trend.

Risperidone response in patients with schizophrenia drives DNA methylation changes in immune and neuronal systems

Background: The choice of efficient antipsychotic therapy for schizophrenia relies on a time-consuming trial-and-error approach, whereas the social and economic burdens of the disease call for faster alternatives. Material & methods: In a search for predictive biomarkers of antipsychotic response, blood methylomes of 28 patients were analyzed before and 4 weeks into risperidone therapy. Results: Several CpGs exhibiting response-specific temporal dynamics were identified in otherwise temporally stable methylomes and noticeable global response-related differences were observed between good and bad responders. These were associated with genes involved in immunity, neurotransmission and neuronal development. Polymorphisms in many of these genes were previously linked with schizophrenia etiology and antipsychotic response. Conclusion: Antipsychotic response seems to be shaped by both stable and medication-induced methylation differences.

Subsampling based variable selection for generalized linear models

A novel variable selection method for low-dimensional generalized linear models is introduced. The new approach called AIC OPTimization via STABility Selection (OPT-STABS) repeatedly subsamples the data, minimizes Akaike's Information Criterion (AIC) over a sequence of nested models for each subsample, and includes in the final model those predictors selected in the minimum AIC model in a large fraction of the subsamples. New methods are also introduced to establish an optimal variable selection cutoff over repeated subsamples. An extensive simulation study examining a variety of proposec variable selection methods shows that, although no single method uniformly outperforms the others in all the scenarios considered, OPT-STABS is consistently among the best-performing methods in most settings while it performs competitively for the rest. This is in contrast to other candidate methods which either have poor performance across the board or exhibit good performance in some settings, but very poor in others. In addition, the asymptotic properties of the OPT-STABS estimator are derived, and its root-n consistency and asymptotic normality are proved. The methods are applied to two datasets involving logistic and Poisson regressions.

Proteomics biomarker discovery for individualized prevention of familial pancreatic cancer using statistical learning

BACKGROUND

The low five-year survival rate of pancreatic ductal adenocarcinoma (PDAC) and the low diagnostic rate of early-stage PDAC via imaging highlight the need to discover novel biomarkers and improve the current screening procedures for early diagnosis. Familial pancreatic cancer (FPC) describes the cases of PDAC that are present in two or more individuals within a circle of first-degree relatives. Using innovative high-throughput proteomics, we were able to quantify the protein profiles of individuals at risk from FPC families in different potential pre-cancer stages. However, the high-dimensional proteomics data structure challenges the use of traditional statistical analysis tools. Hence, we applied advanced statistical learning methods to enhance the analysis and improve the results' interpretability.

METHODS

We applied model-based gradient boosting and adaptive lasso to deal with the small, unbalanced study design via simultaneous variable selection and model fitting. In addition, we used stability selection to identify a stable subset of selected biomarkers and, as a result, obtain even more interpretable results. In each step, we compared the performance of the different analytical pipelines and validated our approaches via simulation scenarios.

RESULTS

In the simulation study, model-based gradient boosting showed a more accurate prediction performance in the small, unbalanced, and high-dimensional datasets than adaptive lasso and could identify more relevant variables. Furthermore, using model-based gradient boosting, we discovered a subset of promising serum biomarkers that may potentially improve the current screening procedure of FPC.

CONCLUSION

Advanced statistical learning methods helped us overcome the shortcomings of an unbalanced study design in a valuable clinical dataset. The discovered serum biomarkers provide us with a clear direction for further investigations and more precise clinical hypotheses regarding the development of FPC and optimal strategies for its early detection.

Spatial heterogeneity of cancer associated protein expression in immunohistochemically stained images as an improved prognostic biomarker

The identification of new tumor biomarkers for patient stratification before therapy, for monitoring of disease progression, and for characterization of tumor biology plays a crucial role in cancer research. The status of these biomarkers is mostly scored manually by a pathologist and such scores typically, do not consider the spatial heterogeneity of the protein's expression in the tissue. Using advanced image analysis methods, marker expression can be determined quantitatively with high accuracy and reproducibility on a per-cell level. To aggregate such per-cell marker expressions on a patient level, the expression values for single cells are usually averaged for the whole tissue. However, averaging neglects the spatial heterogeneity of the marker expression in the tissue. We present two novel approaches for quantitative scoring of spatial marker expression heterogeneity. The first approach is based on a co-occurrence analysis of the marker expression in neighboring cells. The second approach accounts for the local variability of the protein's expression by tiling the tissue with a regular grid and assigning local spatial heterogeneity phenotypes per tile. We apply our novel scores to quantify the spatial expression of four different membrane markers, i.e., HER2, CMET, CD44, and EGFR in immunohistochemically (IHC) stained tissue sections of colorectal cancer patients. We evaluate the prognostic relevance of our spatial scores in this cohort and show that the spatial heterogeneity scores clearly outperform the marker expression average as a prognostic factor (CMET: p-value=0.01 vs. p-value=0.3).

Proteomics of the phase angle: Results from the population-based KORA S4 study

BACKGROUND & AIMS

The phase angle (PhA) measured with bioelectrical impedance analysis is considered to reflect the interrelated components body cell mass and fluid distribution based on technical and physical aspects of the PhA measurement. However, the biomedical meaning of the PhA remains vague. Previous studies mainly assessed associations of the PhA with numerous diseases and health outcomes, but few connected protein markers to the PhA. To broaden our understanding of the biomedical background of the PhA, we aimed to explore a proteomics profile associated with the PhA and related biological factors.

METHODS

The study sample encompassed 1484 participants (725 women and 759 men) aged 55-74 years from the population-based Cooperative Health Research in the Region of Augsburg (KORA) S4 study. Proteomics measurements were performed with a proximity extension assay. We employed boosting with stability selection to establish a set of markers that was strongly associated with the PhA from a group of 233 plasma protein markers. We integrated the selected protein markers into a network and enrichment analysis to identify gene ontology (GO) terms significantly overrepresented for the selected PhA protein markers.

RESULTS

Boosting with stability selection identified seven protein markers that were strongly and independently associated with the PhA: N-terminal prohormone brain natriuretic peptide (NT-proBNP), insulin-like growth factor-binding protein 2 (IGFBP2), adrenomedullin (ADM), myoglobin (MB), matrix metalloproteinase-9 (MMP9), protein-glutamine gamma-glutamyltransferase 2 (TGM2), and fractalkine (CX3CL1) [beta coefficient per 1 standard deviation increase in normalized protein expression values on a log 2 scale (95% confidence interval): -0.12 (-0.15, -0.08), -0.13 (-0.17, -0.09), -0.14 (-0.18, -0.10), 0.10 (0.07, 0.14), 0.07 (0.04, 0.10), 0.08 (0.05, 0.11), -0.06 (-0.10, -0.03), respectively]. According to the enrichment analysis, this protein profile was significantly overrepresented in the following top five GO terms: positive regulation of cell population proliferation (p-value: 1.32E-04), extracellular space (p-value: 1.34E-04), anatomical structure formation involved in morphogenesis (p-value: 2.92E-04), regulation of multicellular organismal development (p-value: 5.72E-04), and metal ion homeostasis (p-value: 8.86E-04).

CONCLUSION

Implementing a proteomics approach, we identified six new protein markers strongly associated with the PhA and confirmed that NT-proBNP is a key PhA marker. The main biological processes that were related to this PhA's protein profile are involved in regulating the amount and growth of cells, reinforcing, from a biomedical perspective, the current technical-based consensus of the PhA to reflect body cell mass.

Identification of shared and disease-specific host gene-microbiome associations across human diseases using multi-omic integration

While gut microbiome and host gene regulation independently contribute to gastrointestinal disorders, it is unclear how the two may interact to influence host pathophysiology. Here we developed a machine learning-based framework to jointly analyse paired host transcriptomic (n = 208) and gut microbiome (n = 208) profiles from colonic mucosal samples of patients with colorectal cancer, inflammatory bowel disease and irritable bowel syndrome. We identified associations between gut microbes and host genes that depict shared as well as disease-specific patterns. We found that a common set of host genes and pathways implicated in gastrointestinal inflammation, gut barrier protection and energy metabolism are associated with disease-specific gut microbes. Additionally, we also found that mucosal gut microbes that have been implicated in all three diseases, such as Streptococcus, are associated with different host pathways in each disease, suggesting that similar microbes can affect host pathophysiology in a disease-specific manner through regulation of different host genes. Our framework can be applied to other diseases for the identification of host gene-microbiome associations that may influence disease outcomes.

Phenotypic Displays of Cholinergic Enzymes Associate With Markers of Inflammation, Neurofibrillary Tangles, and Neurodegeneration in Pre- and Early Symptomatic Dementia Subjects

Background

Cholinergic drugs are the most commonly used drugs for the treatment of Alzheimer’s disease (AD). Therefore, a better understanding of the cholinergic system and its relation to both AD-related biomarkers and cognitive functions is of high importance.

Objectives

To evaluate the relationships of cerebrospinal fluid (CSF) cholinergic enzymes with markers of amyloidosis, neurodegeneration, neurofibrillary tangles, inflammation and performance on verbal episodic memory in a memory clinic cohort.

Methods

In this cross-sectional study, 46 cholinergic drug-free subjects (median age = 71, 54% female, median MMSE = 28) were recruited from an Icelandic memory clinic cohort targeting early stages of cognitive impairment. Enzyme activity of acetylcholinesterase (AChE) and butyrylcholinesterase (BuChE) was measured in CSF as well as levels of amyloid-β_1–42 (Aβ₄₂), phosphorylated tau (P-tau), total-tau (T-tau), neurofilament light (NFL), YKL-40, S100 calcium-binding protein B (S100B), and glial fibrillary acidic protein (GFAP). Verbal episodic memory was assessed with the Rey Auditory Verbal Learning (RAVLT) and Story tests.

Results

No significant relationships were found between CSF Aβ₄₂ levels and AChE or BuChE activity (p > 0.05). In contrast, T-tau (r = 0.46, p = 0.001) and P-tau (r = 0.45, p = 0.002) levels correlated significantly with AChE activity. Although neurodegeneration markers T-tau and NFL did correlate with each other (r = 0.59, p < 0.001), NFL did not correlate with AChE (r = 0.25, p = 0.09) or BuChE (r = 0.27, p = 0.06). Inflammation markers S100B and YKL-40 both correlated significantly with AChE (S100B: r = 0.43, p = 0.003; YKL-40: r = 0.32, p = 0.03) and BuChE (S100B: r = 0.47, p < 0.001; YKL-40: r = 0.38, p = 0.009) activity. A weak correlation was detected between AChE activity and the composite score reflecting verbal episodic memory (r = −0.34, p = 0.02). LASSO regression analyses with a stability approach were performed for the selection of a set of measures best predicting cholinergic activity and verbal episodic memory score. S100B was the predictor with the highest model selection frequency for both AChE (68%) and BuChE (73%) activity. Age (91%) was the most reliable predictor for verbal episodic memory, with selection frequency of both cholinergic enzymes below 10%.

Conclusions

Results indicate a relationship between higher activity of the ACh-degrading cholinergic enzymes with increased neurodegeneration, neurofibrillary tangles and inflammation in the stages of pre- and early symptomatic dementia, independent of CSF Aβ₄₂ levels.

Differential Response of Acidobacteria to Water Content, Soil Type, and Land Use During an Extended Drought in African Savannah Soils

Although climate change is expected to increase the extent of drylands worldwide, the effect of drought on the soil microbiome is still insufficiently understood as for dominant but little characterized phyla like the Acidobacteria. In the present study the active acidobacterial communities of Namibian soils differing in type, physicochemical parameters, and land use were characterized by high-throughput sequencing. Water content, pH, major ions and nutrients were distinct for sandy soils, woodlands or dry agriculture on loamy sands. Soils were repeatedly sampled over a 2-year time period and covered consecutively a strong rainy, a dry, a normal rainy and a weak rainy season. The increasing drought had differential effects on different soils. Linear modeling of the soil water content across all sampling locations and sampling dates revealed that the accumulated precipitation of the preceding season had only a weak, but statistically significant effect, whereas woodland and irrigation exerted a strong positive effect on water content. The decrease in soil water content was accompanied by a pronounced decrease in the fraction of active Acidobacteria (7.9-0.7%) while overall bacterial community size/cell counts remained constant. Notably, the strongest decline in the relative fraction of Acidobacteria was observed after the first cycle of rainy and dry season, rather than after the weakest rainy season at the end of the observation period. Over the 2-year period, also the β-diversity of soil Acidobacteria changed. During the first year this change in composition was related to soil type (loamy sand) and land use (woodland) as explanatory variables. A total of 188 different acidobacterial sequence variants affiliated with the "Acidobacteriia," Blastocatellia, and Vicinamibacteria changed significantly in abundance, suggesting either drought sensitivity or formation of dormant cell forms. Comparative physiological testing of 15 Namibian isolates revealed species-specific and differential responses in viability during long-term continuous desiccation or drying-rewetting cycles. These different responses were not determined by phylogenetic affiliation and provide a first explanation for the effect of drought on soil Acidobacteria. In conclusion, the response of acidobacterial communities to water availability is non-linear, most likely caused by the different physiological adaptations of the different taxa present.

Adaptive step-length selection in gradient boosting for Gaussian location and scale models

Tuning of model-based boosting algorithms relies mainly on the number of iterations, while the step-length is fixed at a predefined value. For complex models with several predictors such as Generalized additive models for location, scale and shape (GAMLSS), imbalanced updates of predictors, where some distribution parameters are updated more frequently than others, can be a problem that prevents some submodels to be appropriately fitted within a limited number of boosting iterations. We propose an approach using adaptive step-length (ASL) determination within a non-cyclical boosting algorithm for Gaussian location and scale models, as an important special case of the wider class of GAMLSS, to prevent such imbalance. Moreover, we discuss properties of the ASL and derive a semi-analytical form of the ASL that avoids manual selection of the search interval and numerical optimization to find the optimal step-length, and consequently improves computational efficiency. We show competitive behavior of the proposed approaches compared to penalized maximum likelihood and boosting with a fixed step-length for Gaussian location and scale models in two simulations and two applications, in particular for cases of large variance and/or more variables than observations. In addition, the underlying concept of the ASL is also applicable to the whole GAMLSS framework and to other models with more than one predictor like zero-inflated count models, and brings up insights into the choice of the reasonable defaults for the step-length in the simpler special case of (Gaussian) additive models.

Deselection of base-learners for statistical boosting-with an application to distributional regression

We present a new procedure for enhanced variable selection for component-wise gradient boosting. Statistical boosting is a computational approach that emerged from machine learning, which allows to fit regression models in the presence of high-dimensional data. Furthermore, the algorithm can lead to data-driven variable selection. In practice, however, the final models typically tend to include too many variables in some situations. This occurs particularly for low-dimensional data (p<n), where we observe a slow overfitting behavior of boosting. As a result, more variables get included into the final model without altering the prediction accuracy. Many of these false positives are incorporated with a small coefficient and therefore have a small impact, but lead to a larger model. We try to overcome this issue by giving the algorithm the chance to deselect base-learners with minor importance. We analyze the impact of the new approach on variable selection and prediction performance in comparison to alternative methods including boosting with earlier stopping as well as twin boosting. We illustrate our approach with data of an ongoing cohort study for chronic kidney disease patients, where the most influential predictors for the health-related quality of life measure are selected in a distributional regression approach based on beta regression.

A review on instance ranking problems in statistical learning

Ranking problems, also known as preference learning problems, define a widely spread class of statistical learning problems with many applications, including fraud detection, document ranking, medicine, chemistry, credit risk screening, image ranking or media memorability. While there already exist reviews concentrating on specific types of ranking problems like label and object ranking problems, there does not yet seem to exist an overview concentrating on instance ranking problems that both includes developments in distinguishing between different types of instance ranking problems as well as careful discussions about their differences and the applicability of the existing ranking algorithms to them. In instance ranking, one explicitly takes the responses into account with the goal to infer a scoring function which directly maps feature vectors to real-valued ranking scores, in contrast to object ranking problems where the ranks are given as preference information with the goal to learn a permutation. In this article, we systematically review different types of instance ranking problems and the corresponding loss functions resp. goodness criteria. We discuss the difficulties when trying to optimize those criteria. As for a detailed and comprehensive overview of existing machine learning techniques to solve such ranking problems, we systematize existing techniques and recapitulate the corresponding optimization problems in a unified notation. We also discuss to which of the instance ranking problems the respective algorithms are tailored and identify their strengths and limitations. Computational aspects and open research problems are also considered.

Predicting Physician Consultations for Low Back Pain Using Claims Data and Population-Based Cohort Data-An Interpretable Machine Learning Approach

(1) Background: Predicting chronic low back pain (LBP) is of clinical and economic interest as LBP leads to disabilities and health service utilization. This study aims to build a competitive and interpretable prediction model; (2) Methods: We used clinical and claims data of 3837 participants of a population-based cohort study to predict future LBP consultations (ICD-10: M40.XX-M54.XX). Best subset selection (BSS) was applied in repeated random samples of training data (75% of data); scoring rules were used to identify the best subset of predictors. The rediction accuracy of BSS was compared to randomforest and support vector machines (SVM) in the validation data (25% of data); (3) Results: The best subset comprised 16 out of 32 predictors. Previous occurrence of LBP increased the odds for future LBP consultations (odds ratio (OR) 6.91 [5.05; 9.45]), while concomitant diseases reduced the odds (1 vs. 0, OR: 0.74 [0.57; 0.98], >1 vs. 0: 0.37 [0.21; 0.67]). The area-under-curve (AUC) of BSS was acceptable (0.78 [0.74; 0.82]) and comparable with SVM (0.78 [0.74; 0.82]) and randomforest (0.79 [0.75; 0.83]); (4) Conclusions: Regarding prediction accuracy, BSS has been considered competitive with established machine-learning approaches. Nonetheless, considerable misclassification is inherent and further refinements are required to improve predictions.

KOMPOS: Connecting Causal Knots in Large Nonlinear Time Series with Non-Parametric Regression Splines

Recovering causality from copious time series data beyond mere correlations has been an important contributing factor in numerous scientific fields. Most existing works assume linearity in the data that may not comply with many real-world scenarios. Moreover, it is usually not sufficient to solely infer the causal relationships. Identifying the correct time delay of cause-effect is extremely vital for further insight and effective policies in inter-disciplinary domains. To bridge this gap, we propose KOMPOS, a novel algorithmic framework that combines a powerful concept from causal discovery of additive noise models with graphical ones. We primarily build our structural causal model from multivariate adaptive regression splines with inherent additive local nonlinearities, which render the underlying causal structure more easily identifiable. In contrast to other methods, our approach is not restricted to Gaussian or non-Gaussian noise due to the non-parametric attribute of the regression method. We conduct extensive experiments on both synthetic and real-world datasets, demonstrating the superiority of the proposed algorithm over existing causal discovery methods, especially for the challenging cases of autocorrelated and non-stationary time series.

Randomized boosting with multivariable base-learners for high-dimensional variable selection and prediction

BACKGROUND

Statistical boosting is a computational approach to select and estimate interpretable prediction models for high-dimensional biomedical data, leading to implicit regularization and variable selection when combined with early stopping. Traditionally, the set of base-learners is fixed for all iterations and consists of simple regression learners including only one predictor variable at a time. Furthermore, the number of iterations is typically tuned by optimizing the predictive performance, leading to models which often include unnecessarily large numbers of noise variables.

RESULTS

We propose three consecutive extensions of classical component-wise gradient boosting. In the first extension, called Subspace Boosting (SubBoost), base-learners can consist of several variables, allowing for multivariable updates in a single iteration. To compensate for the larger flexibility, the ultimate selection of base-learners is based on information criteria leading to an automatic stopping of the algorithm. As the second extension, Random Subspace Boosting (RSubBoost) additionally includes a random preselection of base-learners in each iteration, enabling the scalability to high-dimensional data. In a third extension, called Adaptive Subspace Boosting (AdaSubBoost), an adaptive random preselection of base-learners is considered, focusing on base-learners which have proven to be predictive in previous iterations. Simulation results show that the multivariable updates in the three subspace algorithms are particularly beneficial in cases of high correlations among signal covariates. In several biomedical applications the proposed algorithms tend to yield sparser models than classical statistical boosting, while showing a very competitive predictive performance also compared to penalized regression approaches like the (relaxed) lasso and the elastic net.

CONCLUSIONS

The proposed randomized boosting approaches with multivariable base-learners are promising extensions of statistical boosting, particularly suited for highly-correlated and sparse high-dimensional settings. The incorporated selection of base-learners via information criteria induces automatic stopping of the algorithms, promoting sparser and more interpretable prediction models.

Kidney Allograft Function Is a Confounder of Urine Metabolite Profiles in Kidney Allograft Recipients

Noninvasive biomarkers of kidney allograft status can help minimize the need for standard of care kidney allograft biopsies. Metabolites that are measured in the urine may inform about kidney function and health status, and potentially identify rejection events. To test these hypotheses, we conducted a metabolomics study of biopsy-matched urine cell-free supernatants from kidney allograft recipients who were diagnosed with two major types of acute rejections and no-rejection controls. Non-targeted metabolomics data for 674 metabolites and 577 unidentified molecules, for 192 biopsy-matched urine samples, were analyzed. Univariate and multivariate analyses identified metabolite signatures for kidney allograft rejection. The replicability of a previously developed urine metabolite signature was examined. Our study showed that metabolite profiles can serve as biomarkers for discriminating rejection biopsies from biopsies without rejection features, but also revealed a role of estimated Glomerular Filtration Rate (eGFR) as a major confounder of the metabolite signal.

Proteomic profiling of low muscle and high fat mass: a machine learning approach in the KORA S4/FF4 study

BACKGROUND

The coexistence of low muscle mass and high fat mass, two interrelated conditions strongly associated with declining health status, has been characterized by only a few protein biomarkers. High-throughput proteomics enable concurrent measurement of numerous proteins, facilitating the discovery of potentially new biomarkers.

METHODS

Data derived from the prospective population-based Cooperative Health Research in the Region of Augsburg S4/FF4 cohort study (median follow-up time: 13.5 years) included 1478 participants (756 men and 722 women) aged 55-74 years in the cross-sectional and 608 participants (315 men and 293 women) in the longitudinal analysis. Appendicular skeletal muscle mass (ASMM) and body fat mass index (BFMI) were determined through bioelectrical impedance analysis at baseline and follow-up. At baseline, 233 plasma proteins were measured using proximity extension assay. We implemented boosting with stability selection to enable false positives-controlled variable selection to identify new protein biomarkers of low muscle mass, high fat mass, and their combination. We evaluated prediction models developed based on group least absolute shrinkage and selection operator (lasso) with 100× bootstrapping by cross-validated area under the curve (AUC) to investigate if proteins increase the prediction accuracy on top of classical risk factors.

RESULTS

In the cross-sectional analysis, we identified kallikrein-6, C-C motif chemokine 28 (CCL28), and tissue factor pathway inhibitor as previously unknown biomarkers for muscle mass [association with low ASMM: odds ratio (OR) per 1-SD increase in log2 normalized protein expression values (95% confidence interval (CI)): 1.63 (1.37-1.95), 1.31 (1.14-1.51), 1.24 (1.06-1.45), respectively] and serine protease 27 for fat mass [association with high BFMI: OR (95% CI): 0.73 (0.61-0.86)]. CCL28 and metalloproteinase inhibitor 4 (TIMP4) constituted new biomarkers for the combination of low muscle and high fat mass [association with low ASMM combined with high BFMI: OR (95% CI): 1.32 (1.08-1.61), 1.28 (1.03-1.59), respectively]. Including protein biomarkers selected in ≥90% of group lasso bootstrap iterations on top of classical risk factors improved the performance of models predicting low ASMM, high BFMI, and their combination [delta AUC (95% CI): 0.16 (0.13-0.20), 0.22 (0.18-0.25), 0.12 (0.08-0.17), respectively]. In the longitudinal analysis, N-terminal prohormone brain natriuretic peptide (NT-proBNP) was the only protein selected for loss in ASMM and loss in ASMM combined with gain in BFMI over 14 years [OR (95% CI): 1.40 (1.10-1.77), 1.60 (1.15-2.24), respectively].

CONCLUSIONS

Proteomic profiling revealed CCL28 and TIMP4 as new biomarkers of low muscle mass combined with high fat mass and NT-proBNP as a key biomarker of loss in muscle mass combined with gain in fat mass. Proteomics enable us to accelerate biomarker discoveries in muscle research.

Does diabetes modify the effect of heparin on plasma proteins? - A proteomic search for plasma protein biomarkers for diabetes-related endothelial dysfunction

AIM

Heparin administration affects the concentrations of many plasma proteins through their displacement from the endothelial glycocalyx. A differentiated protein response in diabetes will therefore, at least partly, reflect glycocalyx changes. This study aims at identifying biomarkers of endothelial dysfunction in diabetes by statistical exploration of plasma proteome data for interactions between diabetes status and heparin treatment.

METHODS

Diabetes-by-heparin interactions in relation to protein levels were inspected by regression modelling in plasma proteome data from 497 patients admitted for acute angiography. Analyses were conducted separately for all 273 proteins and as set-based analyses of 44 heparin-relevant proteins identified by gene ontology analysis and 42 heparin-influenced proteins previously reported.

RESULTS

Seventy-five patients had diabetes and 361 received heparin before hospitalization. The proteome-wide analysis displayed no proteins with diabetes-heparin interaction to pass correction for multiple testing. The overall set-based analyses revealed significant association for both protein sets (p-values<2*10^-4), while constraining on opposite directions of effect in diabetics and none-diabetics was insignificant (p-values = 0.11 and 0.17).

CONCLUSIONS

Our plasma proteome-wide interaction approach supports that diabetes influences heparin effects on protein levels, however the direction of effects and individual proteins could not be definitively pinpointed, likely reflecting a complex protein-basis for glycocalyx dysfunction in diabetes.

Where Do We Stand in Regularization for Life Science Studies?

More and more biologists and bioinformaticians turn to machine learning to analyze large amounts of data. In this context, it is crucial to understand which is the most suitable data analysis pipeline for achieving reliable results. This process may be challenging, due to a variety of factors, the most crucial ones being the data type and the general goal of the analysis (e.g., explorative or predictive). Life science data sets require further consideration as they often contain measures with a low signal-to-noise ratio, high-dimensional observations, and relatively few samples. In this complex setting, regularization, which can be defined as the introduction of additional information to solve an ill-posed problem, is the tool of choice to obtain robust models. Different regularization practices may be used depending both on characteristics of the data and of the question asked, and different choices may lead to different results. In this article, we provide a comprehensive description of the impact and importance of regularization techniques in life science studies. In particular, we provide an intuition of what regularization is and of the different ways it can be implemented and exploited. We propose four general life sciences problems in which regularization is fundamental and should be exploited for robustness. For each of these large families of problems, we enumerate different techniques as well as examples and case studies. Lastly, we provide a unified view of how to approach each data type with various regularization techniques.

Psychological experience and coping strategies of patients in the Northeast US delaying care for infertility during the COVID-19 pandemic

BACKGROUND

On March 17, 2020 an expert ASRM task force recommended the temporary suspension of new, non-urgent fertility treatments during an ongoing world-wide pandemic of Covid-19. We surveyed at the time of resumption of fertility care the psychological experience and coping strategies of patients pausing their care due to Covid-19 and examined which factors were associated and predictive of resilience, anxiety, stress and hopefulness.

METHODS

Cross sectional cohort patient survey using an anonymous, self-reported, single time, web-based, HIPPA compliant platform (REDCap). Survey sampled two Northeast academic fertility practices (Yale Medicine Fertility Center in CT and Montefiore's Institute for Reproductive Medicine and Health in NY). Data from multiple choice and open response questions collected demographic, reproductive history, experience and attitudes about Covid-19, prior infertility treatment, sense of hopefulness and stress, coping strategies for mitigating stress and two validated psychological surveys to assess anxiety (six-item short-form State Trait Anxiety Inventory (STAl-6)) and resilience (10-item Connor-Davidson Resilience Scale, (CD-RISC-10).

RESULTS

Seven hundred thirty-four patients were sent invitations to participate. Two hundred fourteen of 734 (29.2%) completed the survey. Patients reported their fertility journey had been delayed a mean of 10 weeks while 60% had been actively trying to conceive > 1.5 years. The top 5 ranked coping skills from a choice of 19 were establishing a daily routine, going outside regularly, exercising, maintaining social connection via phone, social media or Zoom and continuing to work. Having a history of anxiety (p < 0.0001) and having received oral medication as prior infertility treatment (p < 0.0001) were associated with lower resilience. Increased hopefulness about having a child at the time of completing the survey (p < 0.0001) and higher resilience scores (p < 0.0001) were associated with decreased anxiety. Higher reported stress scores (p < 0.0001) were associated with increased anxiety. Multiple multivariate regression showed being non-Hispanic black (p = 0.035) to be predictive of more resilience while variables predictive of less resilience were being a full-time homemaker (p = 0.03), having received oral medication as prior infertility treatment (p = 0.003) and having higher scores on the STAI-6 (< 0.0001).

CONCLUSIONS

Prior to and in anticipation of further pauses in treatment the clinical staff should consider pretreatment screening for psychological distress and provide referral sources. In addition, utilization of a patient centered approach to care should be employed.

Metabolic Predictors of Equine Performance in Endurance Racing

Equine performance in endurance racing depends on the interplay between physiological and metabolic processes. However, there is currently no parameter for estimating the readiness of animals for competition. Our objectives were to provide an in-depth characterization of metabolic consequences of endurance racing and to establish a metabolic performance profile for those animals. We monitored metabolite composition, using a broad non-targeted metabolomics approach, in blood plasma samples from 47 Arabian horses participating in endurance races. The samples were collected before and after the competition and a total of 792 metabolites were measured. We found significant alterations between before and after the race in 417 molecules involved in lipids and amino acid metabolism. Further, even before the race starts, we found metabolic differences between animals who completed the race and those who did not. We identified a set of six metabolite predictors (imidazole propionate, pipecolate, ethylmalonate, 2R-3R-dihydroxybutyrate, β-hydroxy-isovalerate and X-25455) of animal performance in endurance competition; the resulting model had an area under a receiver operating characteristic (AUC) of 0.92 (95% CI: 0.85-0.98). This study provides an in-depth characterization of metabolic alterations driven by endurance races in equines. Furthermore, we showed the feasibility of identifying potential metabolic signatures as predictors of animal performance in endurance competition.

Model selection for inferential models with high dimensional data: synthesis and graphical representation of multiple techniques

Inferential research commonly involves identification of causal factors from within high dimensional data but selection of the 'correct' variables can be problematic. One specific problem is that results vary depending on statistical method employed and it has been argued that triangulation of multiple methods is advantageous to safely identify the correct, important variables. To date, no formal method of triangulation has been reported that incorporates both model stability and coefficient estimates; in this paper we develop an adaptable, straightforward method to achieve this. Six methods of variable selection were evaluated using simulated datasets of different dimensions with known underlying relationships. We used a bootstrap methodology to combine stability matrices across methods and estimate aggregated coefficient distributions. Novel graphical approaches provided a transparent route to visualise and compare results between methods. The proposed aggregated method provides a flexible route to formally triangulate results across any chosen number of variable selection methods and provides a combined result that incorporates uncertainty arising from between-method variability. In these simulated datasets, the combined method generally performed as well or better than the individual methods, with low error rates and clearer demarcation of the true causal variables than for the individual methods.

Network propagation of rare variants in Alzheimer's disease reveals tissue-specific hub genes and communities

State-of-the-art rare variant association testing methods aggregate the contribution of rare variants in biologically relevant genomic regions to boost statistical power. However, testing single genes separately does not consider the complex interaction landscape of genes, nor the downstream effects of non-synonymous variants on protein structure and function. Here we present the NETwork Propagation-based Assessment of Genetic Events (NETPAGE), an integrative approach aimed at investigating the biological pathways through which rare variation results in complex disease phenotypes. We applied NETPAGE to sporadic, late-onset Alzheimer's disease (AD), using whole-genome sequencing from the AD Neuroimaging Initiative (ADNI) cohort, as well as whole-exome sequencing from the AD Sequencing Project (ADSP). NETPAGE is based on network propagation, a framework that models information flow on a graph and simulates the percolation of genetic variation through tissue-specific gene interaction networks. The result of network propagation is a set of smoothed gene scores that can be tested for association with disease status through sparse regression. The application of NETPAGE to AD enabled the identification of a set of connected genes whose smoothed variation profile was robustly associated to case-control status, based on gene interactions in the hippocampus. Additionally, smoothed scores significantly correlated with risk of conversion to AD in Mild Cognitive Impairment (MCI) subjects. Lastly, we investigated tissue-specific transcriptional dysregulation of the core genes in two independent RNA-seq datasets, as well as significant enrichments in terms of gene sets with known connections to AD. We present a framework that enables enhanced genetic association testing for a wide range of traits, diseases, and sample sizes.

RNA sequencing of whole blood in dogs with primary immune-mediated hemolytic anemia (IMHA) reveals novel insights into disease pathogenesis

Immune-mediated hemolytic anemia (IMHA) is a life-threatening autoimmune disorder characterized by a self-mediated attack on circulating red blood cells. The disease occurs naturally in both dogs and humans, but is significantly more prevalent in dogs. Because of its shared features across species, dogs offer a naturally occurring model for studying IMHA in people. In this study, we used RNA sequencing of whole blood from treatment-naïve dogs to study transcriptome-wide changes in gene expression in newly diagnosed animals compared to healthy controls. We found many overexpressed genes in pathways related to neutrophil function, coagulation, and hematopoiesis. In particular, the most highly overexpressed gene in cases was a phospholipase scramblase, which mediates the externalization of phosphatidylserine from the inner to the outer leaflet of cell membranes. This family of genes has been shown to be critically important for programmed cell death of erythrocytes as well as the initiation of the clotting cascade. Unexpectedly, we found marked underexpression of many genes related to lymphocyte function. We also identified groups of genes that are highly associated with the inflammatory response and red blood cell regeneration in affected dogs. We did not find any genes that distinguished dogs that lived vs. those that died at 30 days following diagnosis, nor did we find any relevant genomic signatures of microbial organisms in the blood of affected animals. Future studies are warranted to validate these findings and assess their implication in developing novel therapeutic approaches for dogs and humans with IMHA.

Using machine learning of computerized vocal expression to measure blunted vocal affect and alogia

Negative symptoms are a transdiagnostic feature of serious mental illness (SMI) that can be potentially “digitally phenotyped” using objective vocal analysis. In prior studies, vocal measures show low convergence with clinical ratings, potentially because analysis has used small, constrained acoustic feature sets. We sought to evaluate (1) whether clinically rated blunted vocal affect (BvA)/alogia could be accurately modelled using machine learning (ML) with a large feature set from two separate tasks (i.e., a 20-s “picture” and a 60-s “free-recall” task), (2) whether “Predicted” BvA/alogia (computed from the ML model) are associated with demographics, diagnosis, psychiatric symptoms, and cognitive/social functioning, and (3) which key vocal features are central to BvA/Alogia ratings. Accuracy was high (>90%) and was improved when computed separately by speaking task. ML scores were associated with poor cognitive performance and social functioning and were higher in patients with schizophrenia versus depression or mania diagnoses. However, the features identified as most predictive of BvA/Alogia were generally not considered critical to their operational definitions. Implications for validating and implementing digital phenotyping to reduce SMI burden are discussed.

Association of glial and neuronal degeneration markers with Alzheimer's disease cerebrospinal fluid profile and cognitive functions

BACKGROUND

Neuroinflammation has gained increasing attention as a potential contributing factor in the onset and progression of Alzheimer's disease (AD). The objective of this study was to examine the association of selected cerebrospinal fluid (CSF) inflammatory and neuronal degeneration markers with signature CSF AD profile and cognitive functions among subjects at the symptomatic pre- and early dementia stages.

METHODS

In this cross-sectional study, 52 subjects were selected from an Icelandic memory clinic cohort. Subjects were classified as having AD (n = 28, age = 70, 39% female, Mini-Mental State Examination [MMSE] = 27) or non-AD (n = 24, age = 67, 33% female, MMSE = 28) profile based on the ratio between CSF total-tau (T-tau) and amyloid-β_1-42 (Aβ₄₂) values (cut-off point chosen as 0.52). Novel CSF biomarkers included neurofilament light (NFL), YKL-40, S100 calcium-binding protein B (S100B) and glial fibrillary acidic protein (GFAP), measured with enzyme-linked immunosorbent assays (ELISAs). Subjects underwent neuropsychological assessment for evaluation of different cognitive domains, including verbal episodic memory, non-verbal episodic memory, language, processing speed, and executive functions.

RESULTS

Accuracy coefficient for distinguishing between the two CSF profiles was calculated for each CSF marker and test. Novel CSF markers performed poorly (area under curve [AUC] coefficients ranging from 0.61 to 0.64) compared to tests reflecting verbal episodic memory, which all performed fair (AUC > 70). LASSO regression with a stability approach was applied for the selection of CSF markers and demographic variables predicting performance on each cognitive domain, both among all subjects and only those with a CSF AD profile. Relationships between CSF markers and cognitive domains, where the CSF marker reached stability selection criteria of > 75%, were visualized with scatter plots. Before calculations of corresponding Pearson's correlations coefficients, composite scores for cognitive domains were adjusted for age and education. GFAP correlated with executive functions (r = - 0.37, p = 0.01) overall, while GFAP correlated with processing speed (r = - 0.68, p < 0.001) and NFL with verbal episodic memory (r = - 0.43, p = 0.02) among subjects with a CSF AD profile.

CONCLUSIONS

The novel CSF markers NFL and GFAP show potential as markers for cognitive decline among individuals with core AD pathology at the symptomatic pre- and early stages of dementia.

Population variation in miRNAs and isomiRs and their impact on human immunity to infection

BACKGROUND

MicroRNAs (miRNAs) are key regulators of the immune system, yet their variation and contribution to intra- and inter-population differences in immune responses is poorly characterized.

RESULTS

We generate 977 miRNA-sequencing profiles from primary monocytes from individuals of African and European ancestry following activation of three TLR pathways (TLR4, TLR1/2, and TLR7/8) or infection with influenza A virus. We find that immune activation leads to important modifications in the miRNA and isomiR repertoire, particularly in response to viral challenges. These changes are much weaker than those observed for protein-coding genes, suggesting stronger selective constraints on the miRNA response to stimulation. This is supported by the limited genetic control of miRNA expression variability (miR-QTLs) and the lower occurrence of gene-environment interactions, in stark contrast with eQTLs that are largely context-dependent. We also detect marked differences in miRNA expression between populations, which are mostly driven by non-genetic factors. On average, miR-QTLs explain approximately 60% of population differences in expression of their cognate miRNAs and, in some cases, evolve adaptively, as shown in Europeans for a miRNA-rich cluster on chromosome 14. Finally, integrating miRNA and mRNA data from the same individuals, we provide evidence that the canonical model of miRNA-driven transcript degradation has a minor impact on miRNA-mRNA correlations, which are, in our setting, mainly driven by co-transcription.

CONCLUSION

Together, our results shed new light onto the factors driving miRNA and isomiR diversity at the population level and constitute a useful resource for evaluating their role in host differences of immunity to infection.

Using gradient boosting with stability selection on health insurance claims data to identify disease trajectories in chronic obstructive pulmonary disease

OBJECTIVE

We propose a data-driven method to detect temporal patterns of disease progression in high-dimensional claims data based on gradient boosting with stability selection.

MATERIALS AND METHODS

We identified patients with chronic obstructive pulmonary disease in a German health insurance claims database with 6.5 million individuals and divided them into a group of patients with the highest disease severity and a group of control patients with lower severity. We then used gradient boosting with stability selection to determine variables correlating with a chronic obstructive pulmonary disease diagnosis of highest severity and subsequently model the temporal progression of the disease using the selected variables.

RESULTS

We identified a network of 20 diagnoses (e.g. respiratory failure), medications (e.g. anticholinergic drugs) and procedures associated with a subsequent chronic obstructive pulmonary disease diagnosis of highest severity. Furthermore, the network successfully captured temporal patterns, such as disease progressions from lower to higher severity grades.

DISCUSSION

The temporal trajectories identified by our data-driven approach are compatible with existing knowledge about chronic obstructive pulmonary disease showing that the method can reliably select relevant variables in a high-dimensional context.

CONCLUSION

We provide a generalizable approach for the automatic detection of disease trajectories in claims data. This could help to diagnose diseases early, identify unknown risk factors and optimize treatment plans.

Fast computation of genome-metagenome interaction effects

Motivation

Association studies have been widely used to search for associations between common genetic variants observations and a given phenotype. However, it is now generally accepted that genes and environment must be examined jointly when estimating phenotypic variance. In this work we consider two types of biological markers: genotypic markers, which characterize an observation in terms of inherited genetic information, and metagenomic marker which are related to the environment. Both types of markers are available in their millions and can be used to characterize any observation uniquely.

Objective

Our focus is on detecting interactions between groups of genetic and metagenomic markers in order to gain a better understanding of the complex relationship between environment and genome in the expression of a given phenotype.

Contributions

We propose a novel approach for efficiently detecting interactions between complementary datasets in a high-dimensional setting with a reduced computational cost. The method, named SICOMORE, reduces the dimension of the search space by selecting a subset of supervariables in the two complementary datasets. These supervariables are given by a weighted group structure defined on sets of variables at different scales. A Lasso selection is then applied on each type of supervariable to obtain a subset of potential interactions that will be explored via linear model testing.

Results

We compare SICOMORE with other approaches in simulations, with varying sample sizes, noise, and numbers of true interactions. SICOMORE exhibits convincing results in terms of recall, as well as competitive performances with respect to running time. The method is also used to detect interaction between genomic markers in Medicago truncatula and metagenomic markers in its rhizosphere bacterial community.

Software availability

An R package is available [4], along with its documentation and associated scripts, allowing the reader to reproduce the results presented in the paper.

Stochastic Dispersal Rather Than Deterministic Selection Explains the Spatio-Temporal Distribution of Soil Bacteria in a Temperate Grassland

Spatial and temporal processes shaping microbial communities are inseparably linked but rarely studied together. By Illumina 16S rRNA sequencing, we monitored soil bacteria in 360 stations on a 100 square meter plot distributed across six intra-annual samplings in a rarely managed, temperate grassland. Using a multi-tiered approach, we tested the extent to which stochastic or deterministic processes influenced the composition of local communities. A combination of phylogenetic turnover analysis and null modeling demonstrated that either homogenization by unlimited stochastic dispersal or scenarios, in which neither stochastic processes nor deterministic forces dominated, explained local assembly processes. Thus, the majority of all sampled communities (82%) was rather homogeneous with no significant changes in abundance-weighted composition. However, we detected strong and uniform taxonomic shifts within just nine samples in early summer. Thus, community snapshots sampled from single points in time or space do not necessarily reflect a representative community state. The potential for change despite the overall homogeneity was further demonstrated when the focus shifted to the rare biosphere. Rare OTU turnover, rather than nestedness, characterized abundance-independent β-diversity. Accordingly, boosted generalized additive models encompassing spatial, temporal and environmental variables revealed strong and highly diverse effects of space on OTU abundance, even within the same genus. This pure spatial effect increased with decreasing OTU abundance and frequency, whereas soil moisture – the most important environmental variable – had an opposite effect by impacting abundant OTUs more than the rare ones. These results indicate that – despite considerable oscillation in space and time – the abundant and resident OTUs provide a community backbone that supports much higher β-diversity of a dynamic rare biosphere. Our findings reveal complex interactions among space, time, and environmental filters within bacterial communities in a long-established temperate grassland.

Abnormalities in the genes that encode Large Amino Acid Transporters increase the risk of Autism Spectrum Disorder

Background

Autism spectrum disorder (ASD) is a common neurodevelopmental disorder whose molecular mechanisms are largely unknown. Several studies have shown an association between ASD and abnormalities in the metabolism of amino acids, specifically tryptophan and branched‐chain amino acids (BCAAs).

Methods

Ninety‐seven patients with ASD were screened by Sanger sequencing the genes encoding the heavy (SLC3A2) and light subunits (SLC7A5 and SLC7A8) of the large amino acid transporters (LAT) 1 and 2. LAT1 and 2 are responsible for the transportation of tryptophan and BCAA across the blood–brain barrier and are expressed both in blood and brain. Functional studies were performed employing the Biolog Phenotype Microarray Mammalian (PM‐M) technology to investigate the metabolic profiling in lymphoblastoid cell lines from 43 patients with ASD and 50 controls with particular focus on the amino acid substrates of LATs.

Results

We detected nine likely pathogenic variants in 11 of 97 patients (11.3%): three in SLC3A2, three in SLC7A5, and three in SLC7A8. Six variants of unknown significance were detected in eight patients, two of which also carrying a likely pathogenic variant.

The functional studies showed a consistently reduced utilization of tryptophan, accompanied by evidence of reduced utilization of other large aromatic amino acids (LAAs), either alone or as part of a dipeptide.

Conclusion

Coding variants in the LAT genes were detected in 17 of 97 patients with ASD (17.5%). Metabolic assays indicate that such abnormalities affect the utilization of certain amino acids, particularly tryptophan and other LAAs, with potential consequences on their transport across the blood barrier and their availability during brain development. Therefore, abnormalities in the LAT1 and two transporters are likely associated with an increased risk of developing ASD.

Radiomics in nuclear medicine: robustness, reproducibility, standardization, and how to avoid data analysis traps and replication crisis

Radiomics in nuclear medicine is rapidly expanding. Reproducibility of radiomics studies in multicentre settings is an important criterion for clinical translation. We therefore performed a meta-analysis to investigate reproducibility of radiomics biomarkers in PET imaging and to obtain quantitative information regarding their sensitivity to variations in various imaging and radiomics-related factors as well as their inherent sensitivity. Additionally, we identify and describe data analysis pitfalls that affect the reproducibility and generalizability of radiomics studies. After a systematic literature search, 42 studies were included in the qualitative synthesis, and data from 21 were used for the quantitative meta-analysis. Data concerning measurement agreement and reliability were collected for 21 of 38 different factors associated with image acquisition, reconstruction, segmentation and radiomics-specific processing steps. Variations in voxel size, segmentation and several reconstruction parameters strongly affected reproducibility, but the level of evidence remained weak. Based on the meta-analysis, we also assessed inherent sensitivity to variations of 110 PET image biomarkers. SUV_mean and SUV_max were found to be reliable, whereas image biomarkers based on the neighbourhood grey tone difference matrix and most biomarkers based on the size zone matrix were found to be highly sensitive to variations, and should be used with care in multicentre settings. Lastly, we identify 11 data analysis pitfalls. These pitfalls concern model validation and information leakage during model development, but also relate to reporting and the software used for data analysis. Avoiding such pitfalls is essential for minimizing bias in the results and to enable reproduction and validation of radiomics studies.

Real-time monitoring and model-based prediction of purity and quantity during a chromatographic capture of fibroblast growth factor 2

Process analytical technology combines understanding and control of the process with real‐time monitoring of critical quality and performance attributes. The goal is to ensure the quality of the final product. Currently, chromatographic processes in biopharmaceutical production are predominantly monitored with UV/Vis absorbance and a direct correlation with purity and quantity is limited. In this study, a chromatographic workstation was equipped with additional online sensors, such as multi‐angle light scattering, refractive index, attenuated total reflection Fourier‐transform infrared, and fluorescence spectroscopy. Models to predict quantity, host cell proteins (HCP), and double‐stranded DNA (dsDNA) content simultaneously were developed and exemplified by a cation exchange capture step for fibroblast growth factor 2 expressed in Escherichia coliOnline data and corresponding offline data for product quantity and co‐eluting impurities, such as dsDNA and HCP, were analyzed using boosted structured additive regression. Different sensor combinations were used to achieve the best prediction performance for each quality attribute. Quantity can be adequately predicted by applying a small predictor set of the typical chromatographic workstation sensor signals with a test error of 0.85 mg/ml (range in training data: 0.1–28 mg/ml). For HCP and dsDNA additional fluorescence and/or attenuated total reflection Fourier‐transform infrared spectral information was important to achieve prediction errors of 200 (2–6579 ppm) and 340 ppm (8–3773 ppm), respectively.

Significance Tests for Boosted Location and Scale Models with Linear Base-Learners

Generalized additive models for location scale and shape (GAMLSS) offer very flexible solutions to a wide range of statistical analysis problems, but can be challenging in terms of proper model specification. This complex task can be simplified using regularization techniques such as gradient boosting algorithms, but the estimates derived from such models are shrunken towards zero and it is consequently not straightforward to calculate proper confidence intervals or test statistics. In this article, we propose two strategies to obtain p-values for linear effect estimates for Gaussian location and scale models based on permutation tests and a parametric bootstrap approach. These procedures can provide a solution for one of the remaining problems in the application of gradient boosting algorithms for distributional regression in biostatistical data analyses. Results from extensive simulations indicate that in low-dimensional data both suggested approaches are able to hold the type-I error threshold and provide reasonable test power comparable to the Wald-type test for maximum likelihood inference. In high-dimensional data, when gradient boosting is the only feasible inference for this model class, the power decreases but the type-I error is still under control. In addition, we demonstrate the application of both tests in an epidemiological study to analyse the impact of physical exercise on both average and the stability of the lung function of elderly people in Germany.

Modeling spatiotemporal abundance of mobile wildlife in highly variable environments using boosted GAMLSS hurdle models

Modeling organism distributions from survey data involves numerous statistical challenges, including accounting for zero-inflation, overdispersion, and selection and incorporation of environmental covariates. In environments with high spatial and temporal variability, addressing these challenges often requires numerous assumptions regarding organism distributions and their relationships to biophysical features. These assumptions may limit the resolution or accuracy of predictions resulting from survey-based distribution models. We propose an iterative modeling approach that incorporates a negative binomial hurdle, followed by modeling of the relationship of organism distribution and abundance to environmental covariates using generalized additive models (GAM) and generalized additive models for location, scale, and shape (GAMLSS). Our approach accounts for key features of survey data by separating binary (presence-absence) from count (abundance) data, separately modeling the mean and dispersion of count data, and incorporating selection of appropriate covariates and response functions from a suite of potential covariates while avoiding overfitting. We apply our modeling approach to surveys of sea duck abundance and distribution in Nantucket Sound (Massachusetts, USA), which has been proposed as a location for offshore wind energy development. Our model results highlight the importance of spatiotemporal variation in this system, as well as identifying key habitat features including distance to shore, sediment grain size, and seafloor topographic variation. Our work provides a powerful, flexible, and highly repeatable modeling framework with minimal assumptions that can be broadly applied to the modeling of survey data with high spatiotemporal variability. Applying GAMLSS models to the count portion of survey data allows us to incorporate potential overdispersion, which can dramatically affect model results in highly dynamic systems. Our approach is particularly relevant to systems in which little a priori knowledge is available regarding relationships between organism distributions and biophysical features, since it incorporates simultaneous selection of covariates and their functional relationships with organism responses.