Automated Data Analysis Pipeline Research Articles

SEDA 2024 update: enhancing the SEquence DAtaset builder for seamless integration into automated data analysis pipelines.

The initial version of SEDA assists life science researchers without programming skills with the preparation of DNA and protein sequence FASTA files for multiple bioinformatics applications. However, the initial version of SEDA lacks a command-line interface for more advanced users and does not allow the creation of automated analysis pipelines. The present paper discusses the updates of the new SEDA release, including the addition of a complete command-line interface, new functionalities like gene annotation, a framework for automated pipelines, and improved integration in Linux environments. SEDA is an open-source Java application and can be installed using the different distributions available ( https://www.sing-group.org/seda/download.html ) as well as through a Docker image ( https://hub.docker.com/r/pegi3s/seda ). It is released under a GPL-3.0 license, and its source code is publicly accessible on GitHub ( https://github.com/sing-group/seda ). The software version at the time of submission is archived at Zenodo (version v1.6.0, http://doi.org/10.5281/zenodo.10201605 ).

CytoPipeline and CytoPipelineGUI: a Bioconductor R package suite for building and visualizing automated pre-processing pipelines for flow cytometry data

BackgroundWith the increase of the dimensionality in flow cytometry data over the past years, there is a growing need to replace or complement traditional manual analysis (i.e. iterative 2D gating) with automated data analysis pipelines. A crucial part of these pipelines consists of pre-processing and applying quality control filtering to the raw data, in order to use high quality events in the downstream analyses. This part can in turn be split into a number of elementary steps: signal compensation or unmixing, scale transformation, debris, doublets and dead cells removal, batch effect correction, etc. However, assembling and assessing the pre-processing part can be challenging for a number of reasons. First, each of the involved elementary steps can be implemented using various methods and R packages. Second, the order of the steps can have an impact on the downstream analysis results. Finally, each method typically comes with its specific, non standardized diagnostic and visualizations, making objective comparison difficult for the end user.ResultsHere, we present CytoPipeline and CytoPipelineGUI, two R packages to build, compare and assess pre-processing pipelines for flow cytometry data. To exemplify these new tools, we present the steps involved in designing a pre-processing pipeline on a real life dataset and demonstrate different visual assessment use cases. We also set up a benchmarking comparing two pre-processing pipelines differing by their quality control methods, and show how the package visualization utilities can provide crucial user insight into the obtained benchmark metrics.ConclusionCytoPipeline and CytoPipelineGUI are two Bioconductor R packages that help building, visualizing and assessing pre-processing pipelines for flow cytometry data. They increase productivity during pipeline development and testing, and complement benchmarking tools, by providing user intuitive insight into benchmarking results.

Deep Characterisation of the sn‐Isomer Lipidome Using High‐Throughput Data‐Independent Acquisition and Ozone‐Induced Dissociation**

AbstractIn recent years there has been a significant interest in the development of innovative lipidomics techniques capable of resolving lipid isomers. To date, methods applied to resolving sn‐isomers have resolved only a limited number of species. We report a workflow based on ozone‐induced dissociation for untargeted characterisation of hundreds of sn‐resolved glycerophospholipid isomers from biological extracts in under 20 min, coupled with an automated data analysis pipeline. It provides an order of magnitude increase in the number of sn‐isomer pairs identified as compared to previous reports and reveals that sn‐isomer populations are tightly regulated and significantly different between cell lines. The sensitivity of this method and potential for de novo molecular discovery is further demonstrated by the identification of unexpected lipids containing ultra‐long monounsaturated acyl chains at the sn‐1 position.

Deep Characterisation of the sn-Isomer Lipidome Using High-Throughput Data-Independent Acquisition and Ozone-Induced Dissociation.

In recent years there has been a significant interest in the development of innovative lipidomics techniques capable of resolving lipid isomers. To date, methods applied to resolving sn-isomers have resolved only a limited number of species. We report a workflow based on ozone-induced dissociation for untargeted characterisation of hundreds of sn-resolved glycerophospholipid isomers from biological extracts in under 20 min, coupled with an automated data analysis pipeline. It provides an order of magnitude increase in the number of sn-isomer pairs identified as compared to previous reports and reveals that sn-isomer populations are tightly regulated and significantly different between cell lines. The sensitivity of this method and potential for de novo molecular discovery is further demonstrated by the identification of unexpected lipids containing ultra-long monounsaturated acyl chains at the sn-1 position.

Flow virometry for water-quality assessment: protocol optimization for a model virus and automation of data analysis

Flow virometry (FVM) can support advanced water treatment and reuse by delivering near-real-time information about viral water quality. But maximizing the potential of FVM in water treatment and reuse applications requires protocols to facilitate data validation and interlaboratory comparison—as well as approaches to protocol design to extend the suite of viruses that FVM can feasibly and efficiently monitor. We address these needs herein. First, we optimize a sample-preparation protocol for a model virus using a fractional factorial experimental design. The final protocol for FVM-based detection of T4—an environmentally relevant viral surrogate—blends and improves on existing protocols developed using a traditional pipeline-style optimization approach. Second, we test whether density-based clustering can aid and improve analysis of viral surrogates in complex matrices relative to manual gating. We compare manual gating with results obtained through algorithmic clustering: specifically, by leveraging the OPTICS (Ordering Points to Identify Cluster Structure) ordering algorithm. We demonstrate that OPTICS-assisted clustering can work as well or better than manual gating of FVM data, and can identify features in FVM data difficult to detect through manual gating. We demonstrate our combined sample-preparation and automated data-analysis pipeline on wastewater samples augmented with viral surrogates. We recommend use of this protocol to validate instrument performance prior to and alongside application of FVM on environmental samples. Adoption of a consistent, optimized analytical approach that (i) centers on a widely available, easy-to-use viral target, and (ii) includes automated data analysis will bolster confidence in FVM for microbial water-quality monitoring.

Validation of a suite of ERP and QEEG biomarkers in a pre-competitive, industry-led study in subjects with schizophrenia and healthy volunteers.

Complexity and lack of standardization have mostly limited the use of event-related potentials (ERPs) and quantitative EEG (QEEG) biomarkers in drug development to small early phase trials. We present results from a clinical study on healthy volunteers (HV) and patients with schizophrenia (SZ) that assessed test-retest, group differences, variance, and correlation with functional assessments for ERP and QEEG measures collected at clinical and commercial trial sites with standardized instrumentation and methods, and analyzed through an automated data analysis pipeline. 81 HV and 80 SZ were tested at one of four study sites. Subjects were administered two ERP/EEG testing sessions on separate visits. Sessions included a mismatch negativity paradigm, a 40Hz auditory steady-state response paradigm, an eyes-closed resting state EEG, and an active auditory oddball paradigm. SZ subjects were also tested on the Brief Assessment of Cognition (BAC), Positive and Negative Syndrome Scale (PANSS), and Virtual Reality Functional Capacity Assessment Tool (VRFCAT). Standardized ERP/EEG instrumentation and methods ensured few test failures. The automated data analysis pipeline allowed for near real-time analysis with no human intervention. Test-retest reliability was fair-to-excellent for most of the outcome measures. SZ subjects showed significant deficits in ERP and QEEG measures consistent with published academic literature. A subset of ERP and QEEG measures correlated with functional assessments administered to the SZ subjects. With standardized instrumentation and methods, complex ERP/EEG testing sessions can be reliably performed at clinical and commercial trial sites to produce high-quality data in near real-time.

Mechanistic Understanding of the Discrepancies between Common Peak Picking Algorithms in Liquid Chromatography-Mass Spectrometry-Based Metabolomics.

Inconsistent peak picking outcomes are a critical concern in processing liquid chromatography-mass spectrometry (LC-MS)-based untargeted metabolomics data. This work systematically studied the mechanisms behind the discrepancies among five commonly used peak picking algorithms, including CentWave in XCMS, linear-weighted moving average in MS-DIAL, automated data analysis pipeline (ADAP) in MZmine 2, Savitzky-Golay in El-MAVEN, and FeatureFinderMetabo in OpenMS. We first collected 10 public metabolomics datasets representing various LC-MS analytical conditions. We then incorporated several novel strategies to (i) acquire the optimal peak picking parameters of each algorithm for a fair comparison, (ii) automatically recognize false metabolic features with poor chromatographic peak shapes, and (iii) evaluate the real metabolic features that are missed by the algorithms. By applying these strategies, we compared the true, false, and undetected metabolic features in each data processing outcome. Our results show that linear-weighted moving average consistently outperforms the other peak picking algorithms. To facilitate a mechanistic understanding of the differences, we proposed six peak attributes: ideal slope, sharpness, peak height, mass deviation, peak width, and scan number. We also developed an R program to automatically measure these attributes for detected and undetected true metabolic features. From the results of the 10 datasets, we concluded that four peak attributes, including ideal slope, scan number, peak width, and mass deviation, are critical for the detectability of a peak. For instance, the focus on ideal slope critically hinders the extraction of true metabolic features with low ideal slope scores in linear-weighted moving average, Savitzky-Golay, and ADAP. The relationships between peak picking algorithms and peak attributes were also visualized in a principal component analysis biplot. Overall, the clear comparison and explanation of the differences between peak picking algorithms can lead to the design of better peak picking strategies in the future.

Lateralization of temporal lobe epileptic foci with automated chemical exchange saturation transfer measurements at 3Tesla.

Magnetic Resonance Imaging (MRI) is an indispensable tool for the diagnosis of temporal lobe epilepsy (TLE). However, about 30% of TLE patients show no lesion on structural MRI (sMRI-negative), posing a significant challenge for presurgical evaluation. This study aimed to investigate whether chemical exchange saturation transfer (CEST) MRI at 3 Tesla can lateralize the epileptic focus of TLE and study the metabolic contributors to the CEST signal measured. Forty TLE subjects (16 males and 24 females) were included in this study. An automated data analysis pipeline was established, including segmentation of the hippocampus and amygdala (HA), calculation of four CEST metrics and quantitative relaxation times (T1 and T2), and construction of prediction models by logistic regression. Furthermore, a modified two-stage Bloch-McConnell fitting method was developed to investigate the molecular imaging mechanism of 3T CEST in identifying epileptic foci of TLE. The mean CEST ratio (CESTR) metric within 2.25-3.25ppm in the HA was the most powerful index in predicting seizure laterality, with an area under the receiver-operating characteristic curve (AUC) of 0.84. And, the combination of T2 and CESTR further increased the AUC to 0.92. Amine and guanidinium moieties were the two leading contributors to the CEST contrast between the epileptogenic HA and the normal HA. CEST at 3 Tesla is a powerful modality that can predict seizure laterality with high accuracy. This study can potentially facilitate the clinical translation of CEST MRI in identifying the epileptic foci of TLE or other localization-related epilepsies. National Natural Science Foundation of China, Science Technology Department of Zhejiang Province, and Zhejiang University.

Action potential metrics and automated data analysis pipeline for cardiotoxicity testing using optically mapped hiPSC-derived 3D cardiac microtissues.

Recent advances in human induced pluripotent stem cell (hiPSC)-derived cardiac microtissues provide a unique opportunity for cardiotoxic assessment of pharmaceutical and environmental compounds. Here, we developed a series of automated data processing algorithms to assess changes in action potential (AP) properties for cardiotoxicity testing in 3D engineered cardiac microtissues generated from hiPSC-derived cardiomyocytes (hiPSC-CMs). Purified hiPSC-CMs were mixed with 5-25% human cardiac fibroblasts (hCFs) under scaffold-free conditions and allowed to self-assemble into 3D spherical microtissues in 35-microwell agarose gels. Optical mapping was performed to quantify electrophysiological changes. To increase throughput, AP traces from 4x4 cardiac microtissues were simultaneously acquired with a voltage sensitive dye and a CMOS camera. Individual microtissues showing APs were identified using automated thresholding after Fourier transforming traces. An asymmetric least squares method was used to correct non-uniform background and baseline drift, and the fluorescence was normalized (ΔF/F0). Bilateral filtering was applied to preserve the sharpness of the AP upstroke. AP shape changes under selective ion channel block were characterized using AP metrics including stimulation delay, rise time of AP upstroke, APD30, APD50, APD80, APDmxr (maximum rate change of repolarization), and AP triangulation (APDtri = APDmxr-APD50). We also characterized changes in AP metrics under various ion channel block conditions with multi-class logistic regression and feature extraction using principal component analysis of human AP computer simulations. Simulation results were validated experimentally with selective pharmacological ion channel blockers. In conclusion, this simple and robust automated data analysis pipeline for evaluating key AP metrics provides an excellent in vitro cardiotoxicity testing platform for a wide range of environmental and pharmaceutical compounds.

BioSAXS at European Synchrotron Radiation Facility – Extremely Brilliant Source: BM29 with an upgraded source, detector, robot, sample environment, data collection and analysis software

As part of its Extremely Brilliant Source (EBS) upgrade project, the ESRF's BM29 BioSAXS beamline was subject to a significant upgrade and refurbishment. In addition to the replacement of the beamline's original bending magnet source by a two-pole wiggler, leading to an increase in brilliance by a factor of 60, the sample environment of the beamline was almost completely refurbished: a vacuum-compatible Pilatus3 X 2M with a sensitive area of 253.7 mm × 288 mm and frame rates up to 250 Hz was installed, increasing the active area available and thus the q-scaling of scattering images taken; the sample changer was replaced with an upgraded version, allowing more space for customizable sample environments and the installation of two new sample exposure units; the software associated with the beamline was also renewed. In addition, the layout and functionality of the BSXCuBE3 (BioSAXS Customized Beamline Environment) data acquisition software was redesigned, providing an intuitive `user first' approach for inexperienced users, while at the same time maintaining more powerful options for experienced users and beamline staff. Additional features of BSXCuBE3 are queuing of samples; either consecutive sample changer and/or SEC-SAXS (size-exclusion chromatography small-angle X-ray scattering) experiments, including column equilibration were also implemented. Automatic data processing and analysis are now managed via Dahu, an online server with upstream data reduction, data scaling and azimuthal integration built around PyFAI (Python Fast Azimuthal Integration), and data analysis performed using the open source FreeSAS. The results of this automated data analysis pipeline are displayed in ISPyB/ExiSAXS. The upgraded BM29 has been in operation since the post-EBS restart in September 2020, and here a full description of its new hardware and software characteristics together with examples of data obtained are provided.

High-throughput mechanistic screening of non-equilibrium inhibitors by a fully automated data analysis pipeline in early drug-discovery.

Recent efforts for increasing the success in drug discovery focus on an early, massive, and routine mechanistic and/or kinetic characterization of drug-target engagement as part of a design-make-test-analyze strategy. From an experimental perspective, many mechanistic assays can be translated into a scalable format on automation platforms and thereby enable routine characterization of hundreds or thousands of compounds. However, now the limiting factor to achieve such in-depth characterization at high-throughput becomes the quality-driven data analysis, the sheer scale of which outweighs the time available to the scientific staff of most labs. Therefore, automated analytical workflows are needed to enable such experimental scale-up. We have implemented such a fully automated workflow in Genedata Screener for time-dependent ligand-target binding analysis to characterize non-equilibrium inhibitors. The workflow automates Quality Control (QC) / data modelling and decision-making process in a staged analysis: (1) quality control of raw input data-fluorescence signal-based progress curves - featuring automated rejection of unsuitable measurements; (2) automated model selection - one-step versus two-step binding model - using statistical methods and biological validity rules; (3) result visualization in specific plots and annotated result tables, enabling the scientist to review large result sets efficiently and, at the same time, to rapidly identify and focus on interesting or unusual results; (4) an interactive user interface for immediate adjustment of automated decisions, where necessary. Applying this workflow to first-pass, high-throughput kinetic studies on kinase projects has allowed us to surmount previously rate-limiting manual analysis steps and boost productivity; and is now routinely embedded in a biopharma discovery research process.

Rapid single-molecule characterisation of enzymes involved in nucleic-acid metabolism.

The activity of enzymes is traditionally characterised through bulk-phase biochemical methods that only report on population averages. Single-molecule methods are advantageous in elucidating kinetic and population heterogeneity but are often complicated, time consuming, and lackstatistical power. We present a highly-generalisable and high-throughput single-molecule assay to rapidly characterise proteins involved in DNA metabolism. The assay exclusively relies on changes in total fluorescence intensity of surface-immobilised DNA templates as a result of DNA synthesis, unwinding or digestion. Combined with an automated data-analysis pipeline, our method provides enzymatic activity data of thousands of molecules in less than an hour. We demonstrate our method by characterising three fundamentally different enzyme activities: digestion by the phage λ exonuclease, synthesis by the phage Phi29 polymerase, and unwinding by the E. coli UvrD helicase. We observe the previously unknown activity of the UvrD helicase to remove neutravidin bound to 5'-, but not 3'-ends of biotinylated DNA.

Development of an automated millifluidic platform and data-analysis pipeline for rapid electrochemical corrosion measurements: A pH study on Zn-Ni

We describe the development of a millifluidic based scanning droplet cell platform for rapid and automated corrosion. This system allows for measurement of corrosion properties (e.g., open circuit potential, corrosion current through Tafel and linear polarization resistance measurements, and cyclic voltammograms) on a localized section of a planar sample. Our system is highly automated and flexible, allowing for scripted changing and mixing of solutions and point-to-point motion on the sample. We have also created an automated data analysis pipeline. Here we demonstrate this tool by corroding a plate of electroplated Zn85Ni15 alloy over a range of pH values and correlate our results with XPS measurements and literature.

Memory-Efficient Searching of Gas-Chromatography Mass Spectra Accelerated by Prescreening.

The number of metabolomics studies and spectral libraries for compound annotation (i.e., assigning possible compound identities to a fragmentation spectrum) has been growing steadily in recent years. Accompanying this growth is the number of mass spectra available for searching through those libraries. As the size of spectral libraries grows, accurate and fast compound annotation becomes more challenging. We herein report a prescreening algorithm that was developed to address the speed of spectral search under the constraint of low memory requirements. This prescreening has been incorporated into the Automated Data Analysis Pipeline Spectral Knowledgebase (ADAP-KDB) and can be applied to compound annotation by searching other spectral libraries as well. Performance of the prescreening algorithm was evaluated for different sets of parameters and compared to the original ADAP-KDB spectral search and the MSSearch software. The comparison has demonstrated that the new algorithm is about four-times faster than the original when searching for low-resolution mass spectra, and about as fast as the original when searching for high-resolution mass spectra. However, the new algorithm is still slower than MSSearch due to the relational database design of the former. The new search workflow can be tried out at the ADAP-KDB web portal.

Comparison of Feature Extraction Algorithms for Prediction of Quality Characteristics

Data acquired during production processes often contain redundant and irrelevant features. Thus, to predict quality characteristics from process data, precise feature extraction is essential to sustain a low prediction error and to limit the computational complexity of the deployed machine learning models.Hence, we compare two feature extraction methods: principal component analysis and an autoencoder. Based on an industrial use case, we highlight the advantages of the methods and provide guidance in creating an automated data analysis pipeline for the prediction of quality characteristics. This pipeline is fundamental for other predictive quality applications such as smart experts. Our results favor the principal component analysis for feature extraction, even though it is less expressive than autoencoders.

Scaling Up Single-Cell Proteomics

Single-cell tandem MS has enabled analyzing hundreds of single cells per day and quantifying thousands of proteins across the cells. The broad dissemination of these capabilities can empower the dissection of pathophysiological mechanisms in heterogeneous tissues. Key requirements for achieving this goal include robust protocols performed on widely accessible hardware, robust quality controls, community standards, and automated data analysis pipelines that can pinpoint analytical problems and facilitate their timely resolution. Toward meeting these requirements, this perspective outlines both existing resources and outstanding opportunities, such as parallelization, for catalyzing the wide dissemination of quantitative single-cell proteomics analysis that can be scaled up to tens of thousands of single cells. Indeed, simultaneous parallelization of the analysis of peptides and single cells is a promising approach for multiplicative increase in the speed of performing deep and quantitative single-cell proteomics. The community is ready to begin a virtuous cycle of increased adoption fueling the development of more technology and resources for single-cell proteomics that in turn drive broader adoption, scientific discoveries, and clinical applications.

Real World Data Estimates of Contralateral Nodal Involvement Based on Clinical and Automatically Extracted Treatment Plan Features

To evaluate the association of contralateral nodal involvement (CNI) with automatically defined midline proximity in patients (pts) with head and neck cancer based on treatment planning (TP) real world data (RWD).All available TPs for head and neck cancer pts between 2010 and 2017 were analyzed in an automated data analysis pipeline where the pt's midline was determined objectively using their external contour and discounting distortion due to neck lymphadenopathy via linear modeling. Primary gross tumor volumes (GTVp) and nodal volumes (GTVn) were contoured by treating radiation oncologists. Objective tumor laterality (OTL) was calculated as the difference between the GTVp center of mass and the midline in the R/L direction. OTL accuracy compared to the clinically assigned laterality (when available) was assessed. Midline proximity (MP) was the difference between the medial edge of the GTVp bounding box to the patient midline with positive values representing progressively lateralized tumors and negative values representing tumors crossing midline. Nodal involvement (ipsilateral and contralateral levels 1-5 and retropharyngeal) was determined using the TP nodal target contours labeled by neck level as is standard practice at our institution during treatment planning. The primary endpoint, CNI, was defined based on these TP contours. Volumes of GTVp were calculated from the TP. Clinical variables including HPV status, age, smoking, disease site and subsite were used for model development. The factors predictive of CNI were assessed using logistic regression (LR).Of 2458 pts available for analysis, 2031 had a clinically assigned laterality. The accuracy of automatically assigned laterality was 0.91 when compared with the clinically assigned laterality. Using the automatically assigned laterality (n = 2458), the overall rate of CNI was 45% and varied by disease site and nodal level involved: oropharynx (55.1%, n = 629/1142), larynx (17.6%, n = 124/703), nasopharynx (79.7%, n = 231/290). Multivariable logistic regression (LR) including subsite, HPV status, GTVp, GTVp center of mass relative to midline, GTVp midline proximity, and ipsilateral nodal involvement was performed. In the oropharynx, multivariable LR demonstrated GTVp volume (P < 0.01) and ipsilateral nodes (P < 0.01) were associated with CNI. Neither HPV status or subsite (e.g., base of tongue) was significant on multivariable analysis. In the nasopharynx, GTVp and ipsilateral nodes were significantly associated with CNI on LR. In the larynx, GTVp, ipsilateral nodes, and midline position were all significantly associated with CLI. Nomograms estimating CLI were generated for each disease site based on significant features from the LRs.Automated TP analysis can be used to generate RWD-based predictions of CNI. CNI risk remains a critical radiotherapy endpoint and RWD may help determine CLI risk using combinations of anatomic and clinical features.

Atri-U: assisted image analysis in routine cardiovascular magnetic resonance volumetry of the left atrium

BackgroundArtificial intelligence can assist in cardiac image interpretation. Here, we achieved a substantial reduction in time required to read a cardiovascular magnetic resonance (CMR) study to estimate left atrial volume without compromising accuracy or reliability. Rather than deploying a fully automatic black-box, we propose to incorporate the automated LA volumetry into a human-centric interactive image-analysis process.Methods and resultsAtri-U, an automated data analysis pipeline for long-axis cardiac cine images, computes the atrial volume by: (i) detecting the end-systolic frame, (ii) outlining the endocardial borders of the LA, (iii) localizing the mitral annular hinge points and constructing the longitudinal atrial diameters, equivalent to the usual workup done by clinicians. In every step human interaction is possible, such that the results provided by the algorithm can be accepted, corrected, or re-done from scratch. Atri-U was trained and evaluated retrospectively on a sample of 300 patients and then applied to a consecutive clinical sample of 150 patients with various heart conditions. The agreement of the indexed LA volume between Atri-U and two experts was similar to the inter-rater agreement between clinicians (average overestimation of 0.8 mL/m2 with upper and lower limits of agreement of − 7.5 and 5.8 mL/m2, respectively). An expert cardiologist blinded to the origin of the annotations rated the outputs produced by Atri-U as acceptable in 97% of cases for step (i), 94% for step (ii) and 95% for step (iii), which was slightly lower than the acceptance rate of the outputs produced by a human expert radiologist in the same cases (92%, 100% and 100%, respectively). The assistance of Atri-U lead to an expected reduction in reading time of 66%—from 105 to 34 s, in our in-house clinical setting.ConclusionsOur proposal enables automated calculation of the maximum LA volume approaching human accuracy and precision. The optional user interaction is possible at each processing step. As such, the assisted process sped up the routine CMR workflow by providing accurate, precise, and validated measurement results.

Platform comparison of HTG EdgeSeq and RNA-Seq for gene expression profiling of tumor tissue specimens.

3566 Background: Clinical biomarker studies are often hindered by the availability of tissue specimens of sufficient quality and quantity. While RNA-Seq is often considered the gold standard for measuring mRNA expression levels in cancer tissue, it typically requires multiple formalin-fixed paraffin-embedded (FFPE) tissue sections to extract a sufficient amount of quality RNA for subsequent gene expression profiling analysis. The HTG EdgeSeq technology is a gene expression profiling platform that combines quantitative nuclease protection assay technology with next-generation sequencing detection. Unlike RNA-Seq, the HTG EdgeSeq technology does not require RNA extraction, and can use small amounts of tissue material, typically several mm2, to generate reproducible gene expression profiles. Methods: This study compares the performance of RNA-Seq and HTG's profiling panel, the HTG EdgeSeq Precision Immuno-Oncology Panel (PIP), which is designed to measure expression levels of 1,392 genes focused on tumor/immune interaction. Approximately 1,200 samples from three tumor indications (gastric cancer, colorectal cancer and ovarian cancer) were tested using both technologies. Results: Up to four FFPE slides were used for RNA extraction to support RNA-Seq testing; out of the 1,202 samples processed, 1,099 generated extracted RNA of sufficient quality and quantity (as measured by RNA concentration, RIN score and %DV200) to proceed to sequencing, which resulted in a pass rate of 91.4% for RNA-Seq. The HTG EdgeSeq PIP panel resulted in a pass rate of 97.3% (samples passing QC metrics) when the same 1,200 samples were tested, and required only a single FFPE section owing to the small sample requirement. The t-SNE (a non-linear dimensionality reduction method) analysis of the common 1,358 genes revealed similar clustering of the three cancer indications between the two methods. Correlations across individual genes by sample resulted in the mean Spearman correlation coefficient of 0.73 (95% confidence interval of 0.61 - 0.80). Additionally, gene-wise comparisons across all samples were also evaluated. Conclusions: These data demonstrate that HTG EdgeSeq gene expression panels can be used as a competitive alternative to RNA-Seq, generating equivalent gene expression results, while offering the added benefits of a small sample size requirement, lack of RNA extraction bias, and fully automated data analysis pipeline.

Automated Pipeline for Photovoltaic Module Electroluminescence Image Processing and Degradation Feature Classification

An automated data analysis pipeline is developed to preprocess electroluminescence (EL) module images, and parse the images into individual cells to be used as an input for machine learning algorithms. The dataset used in the study includes EL images of three 60 cell modules from each of five commercial brands at six steps of damp heat exposure, from 500 to 3000 h. Preprocessing of the original raw EL images includes lens distortion correction, filtering, thresholding, convex hull, regression fitting, and perspective transformation to produce planar indexed module and single cell images. Parsing of PV cells from each of the preprocessed 90 EL module images gives us 5400 cell images, which are function of module brand and damp heat exposure step. From the dataset, two unique degradation categories (“cracked” and “corroded”) were observed, while cells that did not degrade were classified as “good.” For supervised machine learning modeling, cell images were sorted into these three classes yielding 3550 images. A training and testing framework with 80:20 sampling ratio was generated using stratified sampling. Three machine learning algorithms (support vector machine, Random Forest, and convolutional neural network) were trained and tuned independently on the training set and then given the test set to predict the scores for each of the three models. Five-fold cross validation was done on training set to tune hyper-parameters of the models. Model prediction scores showed that convolutional neural network outperforms support vector machine and Random Forest for supervised PV cell classification.