scAmpi—A versatile pipeline for single-cell RNA-seq analysis from basics to clinics

Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful technique to decipher tissue composition at the single-cell level and to inform on disease mechanisms, tumor heterogeneity, and the state of the immune microenvironment. Although multiple methods for the computational analysis of scRNA-seq data exist, their application in a clinical setting demands standardized and reproducible workflows, targeted to extract, condense, and display the clinically relevant information. To this end, we designed scAmpi (Single Cell Analysis mRNA pipeline), a workflow that facilitates scRNA-seq analysis from raw read processing to informing on sample composition, clinically relevant gene and pathway alterations, and in silico identification of personalized candidate drug treatments. We demonstrate the value of this workflow for clinical decision making in a molecular tumor board as part of a clinical study.

Single-cell RNA sequencing (scRNA-seq) measures the expression levels across the genes expressed in each single cell. Thus, it is well suited to inform on the cell type composition and the function of cells in different tissues and diseases. However, it is challenging to correctly process and interpret the large amounts of data generated with scRNA-seq. To this end, we have developed an analysis workflow named scAmpi (Single Cell Analysis mRNA pipeline) that starts on the raw sequencing data and performs preprocessing, quality control, and subsequent analysis steps following state-of-the-art recommendations for scRNA-seq processing. The workflow removes low quality cells, assigns a cell type label to each cell, and visualizes the expression of individual genes of interest and functional pathways on the single cells. Moreover, in disease-related analyses scAmpi can link the observed gene expression to potential drug candidates that could be suited to treat the disease.

The structural connectome and motor recovery after stroke: predicting natural recovery

Stroke patients vary considerably in terms of outcomes: some patients present 'natural' recovery proportional to their initial impairment (fitters), while others do not (non-fitters). Thus, a key challenge in stroke rehabilitation is to identify individual recovery potential to make personalized decisions for neuro-rehabilitation, obviating the 'one-size-fits-all' approach. This goal requires (i) the prediction of individual courses of recovery in the acute stage; and (ii) an understanding of underlying neuronal network mechanisms. 'Natural' recovery is especially variable in severely impaired patients, underscoring the special clinical importance of prediction for this subgroup. Fractional anisotropy connectomes based on individual tractography of 92 patients were analysed 2 weeks after stroke (TA) and their changes to 3 months after stroke (TC - TA). Motor impairment was assessed using the Fugl-Meyer Upper Extremity (FMUE) scale. Support vector machine classifiers were trained to separate patients with natural recovery from patients without natural recovery based on their whole-brain structural connectomes and to define their respective underlying network patterns, focusing on severely impaired patients (FMUE < 20). Prediction accuracies were cross-validated internally, in one independent dataset and generalized in two independent datasets. The initial connectome 2 weeks after stroke was capable of segregating fitters from non-fitters, most importantly among severely impaired patients (TA: accuracy = 0.92, precision = 0.93). Secondary analyses studying recovery-relevant network characteristics based on the selected features revealed (i) relevant differences between networks contributing to recovery at 2 weeks and network changes over time (TC - TA); and (ii) network properties specific to severely impaired patients. Important features included the parietofrontal motor network including the intraparietal sulcus, premotor and primary motor cortices and beyond them also attentional, somatosensory or multimodal areas (e.g. the insula), strongly underscoring the importance of whole-brain connectome analyses for better predicting and understanding recovery from stroke. Computational approaches based on structural connectomes allowed the individual prediction of natural recovery 2 weeks after stroke onset, especially in the difficult to predict group of severely impaired patients, and identified the relevant underlying neuronal networks. This information will permit patients to be stratified into different recovery groups in clinical settings and will pave the way towards personalized precision neurorehabilitative treatment.

MGP-AttTCN: An interpretable machine learning model for the prediction of sepsis

With a mortality rate of 5.4 million lives worldwide every year and a healthcare cost of more than 16 billion dollars in the USA alone, sepsis is one of the leading causes of hospital mortality and an increasing concern in the ageing western world. Recently, medical and technological advances have helped re-define the illness criteria of this disease, which is otherwise poorly understood by the medical society. Together with the rise of widely accessible Electronic Health Records, the advances in data mining and complex nonlinear algorithms are a promising avenue for the early detection of sepsis. This work contributes to the research effort in the field of automated sepsis detection with an open-access labelling of the medical MIMIC-III data set. Moreover, we propose MGP-AttTCN: a joint multitask Gaussian Process and attention-based deep learning model to early predict the occurrence of sepsis in an interpretable manner. We show that our model outperforms the current state-of-the-art and present evidence that different labelling heuristics lead to discrepancies in task difficulty. For instance, when predicting sepsis five hours prior to onset on our new realistic labels, our proposed model achieves an area under the ROC curve of 0.660 and an area under the PR curve of 0.483, whereas the (less interpretable) previous state-of-the-art model (MGP-TCN) achieves 0.635 AUROC and 0.460 AUPR and the popular commercial InSight model achieves 0.490 AUROC and 0.359 AUPR.

Primate-restricted KRAB zinc finger proteins and target retrotransposons control gene expression in human neurons

In the first days of embryogenesis, transposable element-embedded regulatory sequences (TEeRS) are silenced by Kruppel-associated box (KRAB) zinc finger proteins (KZFPs). Many TEeRS are subsequently co-opted in transcription networks, but how KZFPs influence this process is largely unknown. We identify ZNF417 and ZNF587 as primate-specific KZFPs repressing HERVK (human endogenous retrovirus K) and SVA (SINE-VNTR-Alu) integrants in human embryonic stem cells (ESCs). Expressed in specific regions of the human developing and adult brain, ZNF417/587 keep controlling TEeRS in ESC-derived neurons and brain organoids, secondarily influencing the differentiation and neurotransmission profile of neurons and preventing the induction of neurotoxic retroviral proteins and an interferon-like response. Thus, evolutionarily recent KZFPs and their TE targets partner up to influence human neuronal differentiation and physiology.

Terbium-161 for PSMA-targeted radionuclide therapy of prostate cancer

Purpose

The prostate-specific membrane antigen (PSMA) has emerged as an interesting target for radionuclide therapy of metastasized castration-resistant prostate cancer (mCRPC). The aim of this study was to investigate ¹⁶¹Tb (T_1/2 = 6.89 days; Eβ^-uperscript>_av = 154 keV) in combination with PSMA-617 as a potentially more effective therapeutic alternative to ¹⁷⁷Lu-PSMA-617, due to the abundant co-emission of conversion and Auger electrons, resulting in an improved absorbed dose profile.

Methods

¹⁶¹Tb was used for the radiolabeling of PSMA-617 at high specific activities up to 100 MBq/nmol. ¹⁶¹Tb-PSMA-617 was tested in vitro and in tumor-bearing mice to confirm equal properties, as previously determined for ¹⁷⁷Lu-PSMA-617. The effects of ¹⁶¹Tb-PSMA-617 and ¹⁷⁷Lu-PSMA-617 on cell viability (MTT assay) and survival (clonogenic assay) were compared in vitro using PSMA-positive PC-3 PIP tumor cells. ¹⁶¹Tb-PSMA-617 was further investigated in therapy studies using PC-3 PIP tumor-bearing mice.

Results

¹⁶¹Tb-PSMA-617 and ¹⁷⁷Lu-PSMA-617 displayed equal in-vitro properties and tissue distribution profiles in tumor-bearing mice. The viability and survival of PC-3 PIP tumor cells were more reduced when exposed to ¹⁶¹Tb-PSMA-617 as compared to the effect obtained with the same activities of ¹⁷⁷Lu-PSMA-617 over the whole investigated concentration range. Treatment of mice with ¹⁶¹Tb-PSMA-617 (5.0 MBq/mouse and 10 MBq/mouse, respectively) resulted in an activity-dependent increase of the median survival (36 vs 65 days) compared to untreated control animals (19 days). Therapy studies to compare the effects of ¹⁶¹Tb-PSMA-617 and ¹⁷⁷Lu-PSMA-617 indicated the anticipated superiority of ¹⁶¹Tb over ¹⁷⁷Lu.

Conclusion

¹⁶¹Tb-PSMA-617 showed superior in-vitro and in-vivo results as compared to ¹⁷⁷Lu-PSMA-617, confirming theoretical dose calculations that indicate an additive therapeutic effect of conversion and Auger electrons in the case of ¹⁶¹Tb. These data warrant more preclinical research for in-depth investigations of the proposed concept, and present a basis for future clinical translation of ¹⁶¹Tb-PSMA-617 for the treatment of mCRPC.

Electronic supplementary material

The online version of this article (10.1007/s00259-019-04345-0) contains supplementary material, which is available to authorized users.