Amplification Bias Research Articles

B-244 DNA Library Amplification to Produce Archived Reference Samples for R&D and Clinical Assay Development, Optimization, and Validation

Abstract Background Rapid advances in clinical diagnostic testing are constrained by finite specimen availability, technical and financial burdens incurred by patient recruitment, and sample collection. Expanding the availability of patient-derived analyte for experimentation could accelerate assay development and regulatory submissions, while reducing demand for additional resource-intensive clinical studies. We developed the AReS (Archived Reference Sample) platform to utilize genomic libraries as templates for PCR amplification. The resulting PCR product, referred to as an AReS library, serves as an alternative sample type from which development, optimization, and validation studies can be iteratively performed. Methods PCR conditions were optimized to maximize yield while minimizing amplification bias, generating approximately 50X more mass than the starting input. To evaluate the AReS process, aliquots of bisulfite converted original DNA libraries were further amplified under optimized conditions to produce AReS libraries. Both the original and AReS libraries were then hybrid captured using Harbinger Health’s proprietary 8.4 Mb panel and sequenced to ≥100X unique median target coverage (MTC) depth. Original libraries were compared to AReS libraries across Picard sequencing metrics by quantifying methylation across our regions of interest and by classification as determined by our cancer yes/no (CYN) determining algorithm. Harbinger Health’s CYN algorithm was developed using a multi-layered logistic regression-based machine learning approach trained on a separate patient cohort and locked prior to being used in this study. In total, we generated 528 AReS libraries from 321 unique patient-derived DNA samples, of which 124 were from patients diagnosed with cancer and 197 from patients with no cancer diagnosis. An additional sub-study was performed on 16 paired original and AReS libraries containing unique molecular identifiers (UMIs). The UMIs allowed for the comparison of individual cfDNA molecules between the two sample types. Results All AReS samples, including both intra- and inter- batch replicates, were highly concordant to the original library. There was no significant difference across sequencing metrics (e.g., conversion efficiency or %CC and MTC). All AReS libraries had similar methylation values to original libraries; Pearson correlation was greater than 98%. Our data also indicated that greater than 97% of AReS libraries were concordant with the original library by our CYN algorithm classification. Read-level UMI analysis identified that approximately 77% of reads were common between the original and AReS libraries. As both the UMI and sequence insert used were identical, these common reads were derived from the same cfDNA molecule. As a frame of reference, sequencing replicates of the original library similarly shared approximately 77% of common reads. In addition, the Pearson correlation of read frequency compared between AReS and original libraries were within 4% difference of the correlation between original library sequencing replicates. These results showed no indication of AReS-derived amplification bias. Conclusion Taken together, the AReS process produces excess libraries that highly reproducible. AReS libraries are functionally and analytically identical to original libraries and can be applied to both research and clinical use.

PGT-A mosaicism based on NGS intermediate copy numbers: is it time to stop reporting them?

Mosaicism represents a genuine real phenomenon, but its high prevalence and undisclosed clinical significance, stress the burden on genetic counseling and the management of PGT-A results. Even though the assumption of mosaicism from NGS intermediate chromosome copy number profiles may represent a reasonable interpretation, other potential technical reasons, including amplification bias, contamination, biopsy technique, or the analysis algorithms,may constitute alternative explanations. Thresholds confining mosaicism ranges are established according to models employing mixtures of normal and abnormal cells with steady conditions of quantity and quality which are unable to reflect the full extent of variability present in a trophectoderm (TE) biopsy specimen. When the concordance of TE with the ICM is considered, mosaic TE biopsies poorly correlate with the chromosomal status of the remaining embryo, displaying mostly ICM aneuploidy in cases of TE high-range mosaics diagnosis and euploidy when mosaicism grade in TE is less than 50% (low-mid range mosaicism), which implies an evident overestimation of mosaicism results. Indeed, a binary classification of NGS profiles that excludes mosaic ranges, including only euploid and aneuploid diagnosis, provides higher specificity and accuracy in identifying abnormal embryos and discarding them. As intermediate copy number profiles do not represent strong evidence of mosaicism but only an inaccurate and misleading assumption, and considering thatno increased risk has been reported in the offspring, until diagnosis specificity is improved and its clinical implications are determined, laboratories should considerlimiting predictions to euploid and aneuploidand stop reporting mosaicism.

Objective Vision

Abstract Convolutional neural networks (CNNs) are a key technology powering the automated technologies of seeing known as computer vision. CNNs have been especially successful in systems that perform object recognition from visual data. This article examines the persistence of a mid-twentieth-century ontology of the digital image in these contemporary technologies. While CNNs are multidimensional, their ontology flattens distinctions between background and foreground, between subjects and objects, and even the relations established among the categories of information used to organize and train these models. This ontology enables the introduction and amplification of bias and troubling correlations and the transfer or slippage of learned associations between humans and objects found in the training image archives. Inspecting and interpreting what CNNs learn and index through their complex architectures can be difficult if not impossible because of how they encode and obfuscate quite human ways of seeing the world and the image repertoires used to train these algorithms that are rife with residues of prior representations.

Predicting acute clinical deterioration with interpretable machine learning to support emergency care decision making

The emergency department (ED) is a fast-paced environment responsible for large volumes of patients with varied disease acuity. Operational pressures on EDs are increasing, which creates the imperative to efficiently identify patients at imminent risk of acute deterioration. The aim of this study is to systematically compare the performance of machine learning algorithms based on logistic regression, gradient boosted decision trees, and support vector machines for predicting imminent clinical deterioration for patients based on cross-sectional patient data extracted from electronic patient records (EPR) at the point of entry to the hospital. We apply state-of-the-art machine learning methods to predict early patient deterioration, based on their first recorded vital signs, observations, laboratory results, and other predictors documented in the EPR. Clinical deterioration in this study is measured by in-hospital mortality and/or admission to critical care. We build on prior work by incorporating interpretable machine learning and fairness-aware modelling, and use a dataset comprising 118, 886 unplanned admissions to Salford Royal Hospital, UK, to systematically compare model variations for predicting mortality and critical care utilisation within 24 hours of admission. We compare model performance to the National Early Warning Score 2 (NEWS2) and yield up to a 0.366 increase in average precision, up to a 21.16% reduction in daily alert rate, and a median 0.599 reduction in differential bias amplification across the protected demographics of age and sex. We use Shapely Additive exPlanations to justify the models’ outputs, verify that the captured data associations align with domain knowledge, and pair predictions with the causal context of each patient’s most influential characteristics. Introducing our modelling to clinical practice has the potential to reduce alert fatigue and identify high-risk patients with a lower NEWS2 that might be missed currently, but further work is needed to trial the models in clinical practice. We encourage future research to follow a systematised approach to data-driven risk modelling to obtain clinically applicable support tools.

Greengenes2 unifies microbial data in a single reference tree

Studies using 16S rRNA and shotgun metagenomics typically yield different results, usually attributed to PCR amplification biases. We introduce Greengenes2, a reference tree that unifies genomic and 16S rRNA databases in a consistent, integrated resource. By inserting sequences into a whole-genome phylogeny, we show that 16S rRNA and shotgun metagenomic data generated from the same samples agree in principal coordinates space, taxonomy and phenotype effect size when analyzed with the same tree.

CRISPR Cas12a-mediated amplification-free digital DNA assay improves the diagnosis and surveillance of Nasopharyngeal carcinoma

Sensitive and accurate cell-free plasma Epstein-Barr virus (EBV) DNA measurement is essential in the routine diagnosis, monitoring and treatment of Nasopharyngeal Carcinoma (NPC). This measurement in commercial and in-house assay are commonly based on real-time quantitative PCR (qPCR) method, which requires reference materials for standardization and lack quantitative precision due to amplification bias or cross-contamination. To address these issues, we developed a CRISPR/Cas12a-mediated amplification-free digital DNA assay, which targets the repetitive sequences of EBV DNA and utilizes the cis-cleavage activity of CRISPR-Cas12a prior to droplet generation. By this mean, more activated Cas12a-crRNA duplexes could be produced for subsequent target detection and counting, thus improving the performance in detecting low EBV DNA load. We demonstrated that it was more robust than conventional qPCR for detecting plasma EBV DNA in a case-control study of 208 participants, especially when the target concentrations were around the diagnostic cut-off value for NPC. More importantly, this assay allowed a more accurate diagnosis of early-stage NPC, with an area under the curve (AUC) of 0.9883 (versus 0.7682 for qPCR). Furthermore, its absolute quantification capability enabled dynamic monitoring of EBV load in NPC patients during initial diagnosis, treatment, and recurrence, thereby potentially improving disease management and prognosis. Taken together, our results demonstrate that this amplification-free digital assay has the potential to be a robust tool to improve the diagnosis and surveillance of NPC.

A primer-independent DNA polymerase-based method for competent whole-genome amplification of intermediate to high GC sequences.

Multiple displacement amplification (MDA) has proven to be a useful technique for obtaining large amounts of DNA from tiny samples in genomics and metagenomics. However, MDA has limitations, such as amplification artifacts and biases that can interfere with subsequent quantitative analysis. To overcome these challenges, alternative methods and engineered DNA polymerase variants have been developed. Here, we present new MDA protocols based on the primer-independent DNA polymerase (piPolB), a replicative-like DNA polymerase endowed with DNA priming and proofreading capacities. These new methods were tested on a genomes mixture containing diverse sequences with high-GC content, followed by deep sequencing. Protocols relying on piPolB as a single enzyme cannot achieve competent amplification due to its limited processivity and the presence of ab initio DNA synthesis. However, an alternative method called piMDA, which combines piPolB with Φ29 DNA polymerase, allows proficient and faithful amplification of the genomes. In addition, the prior denaturation step commonly performed in MDA protocols is dispensable, resulting in a more straightforward protocol. In summary, piMDA outperforms commercial methods in the amplification of genomes andmetagenomes containing high GC sequences and exhibits similar profiling, error rateand variant determination as the non-amplified samples.

AEC-GAN: Adversarial Error Correction GANs for Auto-Regressive Long Time-Series Generation

Large-scale high-quality data is critical for training modern deep neural networks. However, data acquisition can be costly or time-consuming for many time-series applications, thus researchers turn to generative models for generating synthetic time-series data. In particular, recent generative adversarial networks (GANs) have achieved remarkable success in time-series generation. Despite their success, existing GAN models typically generate the sequences in an auto-regressive manner, and we empirically observe that they suffer from severe distribution shifts and bias amplification, especially when generating long sequences. To resolve this problem, we propose Adversarial Error Correction GAN (AEC-GAN), which is capable of dynamically correcting the bias in the past generated data to alleviate the risk of distribution shifts and thus can generate high-quality long sequences. AEC-GAN contains two main innovations: (1) We develop an error correction module to mitigate the bias. In the training phase, we adversarially perturb the realistic time-series data and then optimize this module to reconstruct the original data. In the generation phase, this module can act as an efficient regulator to detect and mitigate the bias. (2) We propose an augmentation method to facilitate GAN's training by introducing adversarial examples. Thus, AEC-GAN can generate high-quality sequences of arbitrary lengths, and the synthetic data can be readily applied to downstream tasks to boost their performance. We conduct extensive experiments on six widely used datasets and three state-of-the-art time-series forecasting models to evaluate the quality of our synthetic time-series data in different lengths and downstream tasks. Both the qualitative and quantitative experimental results demonstrate the superior performance of AEC-GAN over other deep generative models for time-series generation.

An open protocol for modeling T Cell Clonotype repertoires using TCRβ CDR3 sequences

T cell receptor repertoires can be profiled using next generation sequencing (NGS) to measure and monitor adaptive dynamical changes in response to disease and other perturbations. Genomic DNA-based bulk sequencing is cost-effective but necessitates multiplex target amplification using multiple primer pairs with highly variable amplification efficiencies. Here, we utilize an equimolar primer mixture and propose a single statistical normalization step that efficiently corrects for amplification bias post sequencing. Using samples analyzed by both our open protocol and a commercial solution, we show high concordance between bulk clonality metrics. This approach is an inexpensive and open-source alternative to commercial solutions.

EDNA metabarcoding vs metagenomics: an assessment of dietary competition in two estuarine pipefishes

Understanding the dietary preferences of endangered species can be useful in implementing conservation strategies, including habitat restoration, translocation, and captive breeding. Environmental DNA (eDNA) from feces provides a non-invasive method for analysing animal diets. Currently, metabarcoding, a PCR-based approach, is the method of choice for analysing such data. However, this method has limitations, specifically PCR bias, which can result in the overestimation of the importance of certain taxa and failure to detect other taxa because they do not amplify. The present study compared metabarcoding with metagenomics, a PCR-free method, to assess the diversity of prey items in the feces of a critically endangered South African estuarine pipefish, Syngnathus watermeyeri, and its widely distributed congener S. temminckii to investigate potential dietary competition. The metabarcoding results showed a distinct difference between the diets of S. watermeyeri and S. temminckii, with the former mainly consuming calanoid copepods and the latter preferring caridean shrimp. In each case, a single species dominated the sequences generated by metabarcoding. Metagenomics produced more species identifications, and although the same trend was found regarding the preference of S. watermeyeri for copepods and that of S. temminckii for shrimp, this approach identified additional, albeit yet unidentified, copepod species as being important in the diet of S. watermeyeri. We conclude that the lower number of species identified using metabarcoding was most likely a result of amplification bias, resulting in key copepod species missing from the dietary analysis. These findings suggest that metagenomics is not only a useful complementary method for molecular dietary analysis, but may in some cases outperform metabarcoding. However, metagenomics is even more strongly affected by the lack of reference sequences than is metabarcoding, as the majority of sequences originate from genomic regions that have not yet been sequenced for the putative prey species in question.

Bias Analysis in Healthcare Time Series (BAHT) Decision Support Systems from Meta Data.

One of the hindrances in the widespread acceptance of deep learning-based decision support systems in healthcare is bias. Bias in its many forms occurs in the datasets used to train and test deep learning models and is amplified when deployed in the real world, leading to challenges such as model drift. Recent advancements in the field of deep learning have led to the deployment of deployable automated healthcare diagnosis decision support systems at hospitals as well as tele-medicine through IoT devices. Research has been focused primarily on the development and improvement of these systems leaving a gap in the analysis of the fairness. The domain of FAccT ML (fairness, accountability, and transparency) accounts for the analysis of these deployable machine learning systems. In this work, we present a framework for bias analysis in healthcare time series (BAHT) signals such as electrocardiogram (ECG) and electroencephalogram (EEG). BAHT provides a graphical interpretive analysis of bias in the training, testing datasets in terms of protected variables, and analysis of bias amplification by the trained supervised learning model for time series healthcare decision support systems. We thoroughly investigate three prominent time series ECG and EEG healthcare datasets used for model training and research. We show the extensive presence of bias in the datasets leads to potentially biased or unfair machine-learning models. Our experiments also demonstrate the amplification of identified bias with an observed maximum of 66.66%. We investigate the effect of model drift due to unanalyzed bias in datasets and algorithms. Bias mitigation though prudent is a nascent area of research. We present experiments and analyze the most prevalently accepted bias mitigation strategies of under-sampling, oversampling, and the use of synthetic data for balancing the dataset through augmentation. It is important that healthcare models, datasets, and bias mitigation strategies should be properly analyzed for a fair unbiased delivery of service.

CRISPR-Analytics (CRISPR-A): A platform for precise analytics and simulations for gene editing.

Gene editing characterization with currently available tools does not always give precise relative proportions among the different types of gene edits present in an edited bulk of cells. We have developed CRISPR-Analytics, CRISPR-A, which is a comprehensive and versatile genome editing web application tool and a nextflow pipeline to give support to gene editing experimental design and analysis. CRISPR-A provides a robust gene editing analysis pipeline composed of data analysis tools and simulation. It achieves higher accuracy than current tools and expands the functionality. The analysis includes mock-based noise correction, spike-in calibrated amplification bias reduction, and advanced interactive graphics. This expanded robustness makes this tool ideal for analyzing highly sensitive cases such as clinical samples or experiments with low editing efficiencies. It also provides an assessment of experimental design through the simulation of gene editing results. Therefore, CRISPR-A is ideal to support multiple kinds of experiments such as double-stranded DNA break-based engineering, base editing (BE), primer editing (PE), and homology-directed repair (HDR), without the need of specifying the used experimental approach.

Minimizing amplification bias during reverse transcription for in vitro selections.

Systematic evolution of ligands through exponential enrichment (SELEX) is widely used to identify functional nucleic acids, such as aptamers and ribozymes. Ideally, selective pressure drives the enrichment of sequences that display the function of interest (binding, catalysis, etc.). However, amplification biases from reverse transcription can overwhelm this enrichment and leave some functional sequences at a disadvantage, with cumulative effects across multiple rounds of selection. Libraries that are designed to include structural scaffolds can improve selection outcomes by sampling sequence space more strategically, but they are also susceptible to such amplification biases, particularly during reverse transcription. Therefore, we tested five reverse transcriptases (RTs)-ImProm-II, Marathon RT (MaRT), TGIRT-III, SuperScript IV (SSIV), and BST 3.0 DNA polymerase (BST)-to determine which enzymes introduced the least bias. We directly compared cDNA yield and processivity for these enzymes on RNA templates with varying degrees of structure under various reaction conditions. In these analyses, BST exhibited excellent processivity, generated large quantities of the full-length cDNA product, displayed little bias among templates with varying structure and sequence, and performed well on long, highly structured viral RNAs. Additionally, six RNA libraries containing either strong, moderate, or no incorporated structural elements were pooled and competed head-to-head in six rounds of an amplification-only selection without external selective pressure using either SSIV, ImProm-II, or BST during reverse transcription. High-throughput sequencing established that BST maintained the most neutral enrichment values, indicating low interlibrary bias over the course of six rounds, relative to SSIV and ImProm-II, and it introduced minimal mutational bias.

Signal and noise in metabarcoding data.

Metabarcoding is a powerful molecular tool for simultaneously surveying hundreds to thousands of species from a single sample, underpinning microbiome and environmental DNA (eDNA) methods. Deriving quantitative estimates of underlying biological communities from metabarcoding is critical for enhancing the utility of such approaches for health and conservation. Recent work has demonstrated that correcting for amplification biases in genetic metabarcoding data can yield quantitative estimates of template DNA concentrations. However, a major source of uncertainty in metabarcoding data stems from non-detections across technical PCR replicates where one replicate fails to detect a species observed in other replicates. Such non-detections are a special case of variability among technical replicates in metabarcoding data. While many sampling and amplification processes underlie observed variation in metabarcoding data, understanding the causes of non-detections is an important step in distinguishing signal from noise in metabarcoding studies. Here, we use both simulated and empirical data to 1) suggest how non-detections may arise in metabarcoding data, 2) outline steps to recognize uninformative data in practice, and 3) identify the conditions under which amplicon sequence data can reliably detect underlying biological signals. We show with both simulations and empirical data that, for a given species, the rate of non-detections among technical replicates is a function of both the template DNA concentration and species-specific amplification efficiency. Consequently, we conclude metabarcoding datasets are strongly affected by (1) deterministic amplification biases during PCR and (2) stochastic sampling of amplicons during sequencing-both of which we can model-but also by (3) stochastic sampling of rare molecules prior to PCR, which remains a frontier for quantitative metabarcoding. Our results highlight the importance of estimating species-specific amplification efficiencies and critically evaluating patterns of non-detection in metabarcoding datasets to better distinguish environmental signal from the noise inherent in molecular detections of rare targets.

Digital microfluidics-based digital counting of single-cell copy number variation (dd-scCNV Seq)

Single-cell copy number variations (CNVs), major dynamic changes in humans, result in differential levels of gene expression and account for adaptive traits or underlying disease. Single-cell sequencing is needed to reveal these CNVs but has been hindered by single-cell whole-genome amplification (scWGA) bias, leading to inaccurate gene copy number counting. In addition, most of the current scWGA methods are labor intensive, time-consuming, and expensive with limited wide application. Here, we report a unique single-cell whole-genome library preparation approach based on digital microfluidics for digital counting of single-cell Copy Number Variation (dd-scCNV Seq). dd-scCNV Seq directly fragments the original single-cell DNA and uses these fragments as templates for amplification. These reduplicative fragments can be filtered computationally to generate the original partitioned unique identified fragments, thereby enabling digital counting of copy number variation. dd-scCNV Seq showed an increase in uniformity in the single-molecule data, leading to more accurate CNV patterns compared to other methods with low-depth sequencing. Benefiting from digital microfluidics, dd-scCNV Seq allows automated liquid handling, precise single-cell isolation, and high-efficiency and low-cost genome library preparation. dd-scCNV Seq will accelerate biological discovery by enabling accurate profiling of copy number variations at single-cell resolution.

A Metagenomic and Amplicon Sequencing Combined Approach Reveals the Best Primers to Study Marine Aerobic Anoxygenic Phototrophs

Studies based on protein-coding genes are essential to describe the diversity within bacterial functional groups. In the case of aerobic anoxygenic phototrophic (AAP) bacteria, the pufM gene has been established as the genetic marker for this particular functional group, although available primers are known to have amplification biases. We review here the existing primers for pufM gene amplification, design new ones, and evaluate their phylogenetic coverage. We then use samples from contrasting marine environments to evaluate their performance. By comparing the taxonomic composition of communities retrieved with metagenomics and with different amplicon approaches, we show that the commonly used PCR primers are biased towards the Gammaproteobacteria phylum and some Alphaproteobacteria clades. The metagenomic approach, as well as the use of other combinations of the existing and newly designed primers, show that these groups are in fact less abundant than previously observed, and that a great proportion of pufM sequences are affiliated to uncultured representatives, particularly in the open ocean. Altogether, the framework developed here becomes a better alternative for future studies based on the pufM gene and, additionally, serves as a reference for primer evaluation of other functional genes.

DNA storage in thermoresponsive microcapsules for repeated random multiplexed data access.

DNA has emerged as an attractive medium for archival data storage due to its durability and high information density. Scalable parallel random access to information is a desirable property of any storage system. For DNA-based storage systems, however, this still needs to be robustly established. Here we report on a thermoconfined polymerase chain reaction, which enables multiplexed, repeated random access to compartmentalized DNA files. The strategy is based on localizing biotin-functionalized oligonucleotides inside thermoresponsive, semipermeable microcapsules. At low temperatures, microcapsules are permeable to enzymes, primers and amplified products, whereas at high temperatures, membrane collapse prevents molecular crosstalk during amplification. Our data show that the platform outperforms non-compartmentalized DNA storage compared with repeated random access and reduces amplification bias tenfold during multiplex polymerase chain reaction. Using fluorescent sorting, we also demonstrate sample pooling and data retrieval by microcapsule barcoding. Therefore, the thermoresponsive microcapsule technology offers a scalable, sequence-agnostic approach for repeated random access to archival DNA files.

Invited Perspective: Challenges in Evaluating the Effect of Per- and Polyfluoroalkyl Substance Mixtures on Polycystic Ovarian Syndrome

Invited Perspective: Challenges in Evaluating the Effect of Per- and Polyfluoroalkyl Substance Mixtures on Polycystic Ovarian Syndrome

Abstract A23: Less is more: Ultra-low input sequencing allows for evaluation of stem cells as a predictor of post-transplant relapse in the myelodysplastic syndromes

Abstract Malignant hematopoietic stem cells (HSCs) initiate the myelodysplastic syndromes (MDS) and drive treatment failure through their resistance to all non-transplant therapies. As a result, allogeneic stem cell transplantation (allo-SCT) is the only curative therapy for MDS. However, relapse remains the leading cause of treatment failure after allo-SCT. Improved tools to predict relapse in this setting are necessary to allow for pre-emptive therapeutic interventions, which may be more effective when the burden of disease is still low. Given prior evidence that HSCs serve as the functional unit of resistance to non-transplant therapies, we asked whether relapse after transplant originates from MDS HSCs, and whether their persistence can predict relapse. To overcome the limiting number of HSCs available in clinical bone marrow (BM) samples, we adapted a workflow developed for assessment of cell-free DNA and utilized a hybridization capture-based targeted sequencing panel covering 260 myeloid malignancy-associated genes. Unique dual index primers (UDIs) were utilized to tag input DNA, improving PCR duplicate identification, which overcome issues with amplification bias typically faced by low-input sequencing. This ultimately allowed us to consistently capture evaluable read depths from as few as 10 sorted cells. We assembled an initial cohort of 21 patients who underwent allo-SCT for MDS or secondary acute myeloid leukemia. Inclusion criteria were follow-up time greater than 1.5 years (or relapse in this timeframe), BM sampling prior to Day+120 post-transplant and relapse, and biospecimens allowing for sorting of greater than 10 HSCs. We sequenced sorted CD34+CD38-CD45RA- HSCs and multipotent progenitors (MPPs), CD34+CD38- CD45RA+ lymphoid-primed multipotent progenitors (LMPPs), and CD34+CD38+ progenitors, as well as bulk BM. With a median follow-up of two years, 9 of the 21 patients relapsed, with an average time to relapse of 362 days. Detection of disease mutations in HSCs/MPPs in any BM at D+30-D+120 post-transplant was 100% specific and 88% sensitive for relapse, showing improvements in both metrics over bulk BM assessment (22% sensitive and 66% specific), despite far lower cell input and read depth. Average time from detection to relapse was 6.9 months. In patients destined to relapse, disease mutations are often present in HSCs at a variant allele frequency of greater than 45%, while simultaneously undetectable on high-depth sequencing of bulk cells (>400x coverage). While even higher depth error-corrected sequencing bulk BM could likely detect such mutations, it would require >20x the sequencing depth, and could not identify the involved cell populations. In conclusion, we show for the first time that relapse of MDS after allo-SCT is driven by failure to eradicate MDS HSCs. Detection of MDS HSCs early post-transplant is highly predictive for relapse and opens new doors for pre-emptive interventions to prevent relapse. These findings are being validated in an independent cohort of 46 patients and updated data will be presented at the meeting. Citation Format: Benjamin Kroger, Yi Huang, Prapti Patel, Aimaz Afrough, Gurbakhash Kaur, Larry Anderson, Farrukh Awan, Praveen Ramakrishnan, Kunal Verma, Russell Vittrup, Robert H Collins, Madhuri Vusirikala, Yazan F Madanat, David Sallman, Carmelo Gurnari, Jaroslaw Maciejewski, Roni Tamari, Stephen Chung. Less is more: Ultra-low input sequencing allows for evaluation of stem cells as a predictor of post-transplant relapse in the myelodysplastic syndromes [abstract]. In: Proceedings of the AACR Special Conference: Acute Myeloid Leukemia and Myelodysplastic Syndrome; 2023 Jan 23-25; Austin, TX. Philadelphia (PA): AACR; Blood Cancer Discov 2023;4(3_Suppl):Abstract nr A23.

Targeted DNA methylation analysis and prediction of smoking habits in blood based on massively parallel sequencing

Tobacco smoking is a frequent habit sustained by > 1.3 billion people in 2020 and the leading preventable factor for health risk and premature mortality worldwide. In the forensic context, predicting smoking habits from biological samples may allow broadening DNA phenotyping. In this study, we aimed to implement previously published smoking habit classification models based on blood DNA methylation at 13 CpGs. First, we developed a matching lab tool based on bisulfite conversion and multiplex PCR followed by amplification-free library preparation and targeted paired-end massively parallel sequencing (MPS). Analysis of six technical duplicates revealed high reproducibility of methylation measurements (Pearson correlation of 0.983). Artificially methylated standards uncovered marker-specific amplification bias, which we corrected via bi-exponential models. We then applied our MPS tool to 232 blood samples from Europeans of a wide age range, of which 90 were current, 71 former and 71 never smokers. On average, we obtained 189,000 reads/sample and 15,000 reads/CpG, without marker drop-out. Methylation distributions per smoking category roughly corresponded to previous microarray analysis, showcasing large inter-individual variation but with technology-driven bias. Methylation at 11 out of 13 smoking-CpGs correlated with daily cigarettes in current smokers, while solely one was weakly correlated with time since cessation in former smokers. Interestingly, eight smoking-CpGs correlated with age, and one displayed weak but significant sex-associated methylation differences. Using bias-uncorrected MPS data, smoking habits were relatively accurately predicted using both two- (current/non-current) and three- (never/former/current) category model, but bias correction resulted in worse prediction performance for both models. Finally, to account for technology-driven variation, we built new, joint models with inter-technology corrections, which resulted in improved prediction results for both models, with or without PCR bias correction (e.g. MPS cross-validation F1-score > 0.8; 2-categories). Overall, our novel assay takes us one step closer towards the forensic application of viable smoking habit prediction from blood traces. However, future research is needed towards forensically validating the assay, especially in terms of sensitivity. We also need to further shed light on the employed biomarkers, particularly on the mechanistics, tissue specificity and putative confounders of smoking epigenetic signatures.