Biology subjects

Bal, A.

Publications and source records attributed to Bal, A..

13 recordsLinked to original sources

bioRxiv2026

Quantifying Evidence for Competing Biomedical Hypotheses using Large Language Models and Bayesian Analysis

Science fundamentally depends on the generation and testing of hypotheses, many of them controversial. An explosion in scientific literature has made evaluating hypotheses even within a domain a problem of scale, and risks slowing an already extensive consensus-building process. While this challenge has prompted interest in automated hypothesis evaluation tools, existing methods have not yet proven effective for comparing hypotheses. Here, we introduce KM-GPT-DCH, an algorithm that combines co-occurrence methods with large language models (LLMs) to develop a transparent and reproducible literature-based algorithm to compare controversial hypotheses using a structured scoring approach with Bayesian methods to estimate confidence. When testing the algorithm on historical controversial hypotheses previously decided, KM-GPT-DCH chooses the correct hypothesis with high confidence several years before the scientific community or public do so. We further apply the algorithm to compare twenty unresolved controversial hypothesis pairs providing guidance for future research. The method can help researchers and the public to evaluate biomedical hypotheses such as "Is it more likely that monoamine deficiency or inflammation causes depression?" It can also be used to assess and visualize historical trends in the scientific literature. A web-based implementation of the algorithm is freely available at https://skim.morgridge.org.

bioinformatics↗

bioRxiv2025

SKiM-GPT: Combining Biomedical Literature-Based Discovery with Large Language Model Hypothesis Evaluation

BackgroundGenerating and testing hypotheses is a critical aspect of biomedical science. Typically, researchers generate hypotheses by carefully analyzing available information and making logical connections, which are then tested. The accelerating growth of biomedical literature makes it increasingly difficult to keep pace with connections between biological entities emerging across biomedical research. Recently developed automated means of generating hypotheses can generate many more hypotheses than can be easily tested. One such approach involves literature based discovery (LBD) systems such as Serial KinderMiner (SKiM), which surfaces putative A B C links derived from term co occurrence. However, LBD systems leave three critical gaps: (i) they find statistical associations, not biological relationships; (ii) they can produce false positive leads; and (iii) they do not assess agreement with a hypothesis in question. As a result, LBD search results often require costly manual curation to be of practical utility to the researcher. Large language models (LLMs) have the potential to automate much of this curation step, but standalone LLMs are hampered by hallucinations, lack of transparency in information sources, and the inability to reference data not included in the training corpus. ResultsWe introduce SKiM-GPT, a retrieval-augmented generation (RAG) system that combines SKiMs co-occurrence search and retrieval with frontier LLMs to evaluate user- defined hypotheses. For every chosen A-B-C SKiM hit, SKiM-GPT retrieves appropriate PubMed abstract texts, filters out irrelevant abstracts with a fine-tuned relevance model, and prompts an LLM to evaluate the users hypothesis, given the relevant abstracts. Importantly, the SKiM-GPT system is transparent and human-verifiable: it displays the retrieved abstracts, the hypothesis score, and a justification for the score grounded in the texts and written in natural language. On a benchmark consisting of 14 disease-gene-drug hypotheses, SKiM-GPT achieves strong ordinal agreement with four expert biologists (Cohens {kappa} = 0.84), demonstrating its ability to replicate expert judgment. ConclusionsSKiM-GPT is open-source (https://github.com/stewart-lab/skimgpt) and available through a web interface (https://skim.morgridge.org), enabling both wet-lab and computational researchers to systematically and efficiently evaluate biomedical hypotheses at scale.

bioinformatics↗

bioRxiv2025

Social Exclusion Amplifies Behavioral Responses to Physical Pain via Insular Neuromodulation

The "Pain Overlap Theory" (1) proposes that the experience of social pain overlaps with and amplifies the experience of physical pain by sharing parts of the same underlying processing systems (2-6). In humans, the insular cortex has been implicated in this overlap of physical and social pain, but a mechanistic link has not been made (2,4,5,7-9). To determine whether social pain can subsequently impact responses to nociceptive stimuli via convergent electrical signals (spikes) or convergent chemical signals (neuromodulators), we designed a novel Social Exclusion paradigm termed the Fear of Missing Out (FOMO) Task which facilitates a mechanistic investigation in mice. We found that socially-excluded mice display more severe responses to physical pain, disrupted valence encoding, and impaired neural representations of nociceptive stimuli. We performed a systematic biosensor panel and found that endocannabinoid and oxytocin signaling in the insular cortex have opposing responses during trials where mice were attending or not attending to the Social Exclusion events respectively, demonstrating distinct neuromodulatory substrates that underpin different states of Social Exclusion. We also found that intra-insular blockade of oxytocin signaling increased the response to physical pain following Social Exclusion. Together these findings suggest Social Exclusion effectively alters physical pain perception using neuromodulatory signaling in the insular cortex.

neuroscience↗

bioRxiv2023

Mesostriatal dopamine is sensitive to specific cue-reward contingencies

Learning causal relationships relies on understanding how often one event precedes another. To gain an understanding of how dopamine neuron activity and neurotransmitter release change when a retrospective relationship is degraded for a specific pair of events, we used outcome-selective Pavlovian contingency degradation in rats. Two cues were paired with distinct food rewards, one of which was also delivered in the absence of either cue. Conditioned approach was attenuated for the cue-reward contingency that was degraded. Dopamine neuron activity in the midbrain and dopamine release in the ventral striatum showed a profile of changes in cue- and reward-evoked responding that was not easily explained by a standard reinforcement learning model. An alternative model based on learning causal relationships was better able to capture evoked dopamine responses during contingency degradation, as well as conditioned behavior following optogenetic manipulations of dopamine during noncontingent rewards. Our results suggest that mesostriatal dopamine encodes the contingencies between meaningful events during learning.

neuroscience↗

bioRxiv2020

AlphaTracker: A Multi-Animal Tracking and Behavioral Analysis Tool

The advancement of behavioral analysis in neuroscience has been aided by the development of computational tools1,2. Specifically, computer vision algorithms have emerged as a powerful tool to elevate behavioral research3,4. Yet fully automatic analysis of social behavior remains challenging in two ways. First, existing tools to track and analyze behavior often focus on single animals, not multiple, interacting animals. Second, many available tools are not developed for novice users and require programming experience to run. Here, we unveil a computer vision pipeline called AlphaTracker, which requires minimal hardware requirements and produces reliable tracking of multiple unmarked animals. An easy-to-use user interface further enables manual inspection and curation of results. We demonstrate the practical, real-time advantages of AlphaTracker through the study of multiple, socially-interacting mice.

animal behavior and cognition↗

bioRxiv2020

Characterization of SARS-CoV-2 ORF6 deletion variants detected in a nosocomial cluster during routine genomic surveillance, Lyon, France

Through routine genomic surveillance of the novel SARS-CoV-2 virus (n=229 whole genome sequences), 2 different frameshifting deletions were newly detected in the open reading frame (ORF) 6, starting at the same position (27267). While the 26-nucleotide deletion variant was only found in one sample in March 2020, the 34-nucleotide deletion variant was found within a single geriatric hospital unit in 5/9 patients sequenced and one health care worker with samples collected between April 2nd and 9th, 2020. Both the presence of the 34-nucleotide deletion variant limited to this unit and the clustering of the corresponding whole genome sequences by phylogeny analysis strongly suggested a nosocomial transmission between patients. Interestingly, prolonged viral excretion of the 34-nucleotide deletion variant was identified in a stool sample 14 days after initial diagnosis for one patient. Clinical data revealed no significant difference in disease severity between patients harboring the wild-type or the 34-nucleotide deletion variants. The in vitro infection of the two deletion variants on primate endothelial kidney cells (BGM) and human lung adenocarcinoma cells (Calu-3) yielded comparable replication kinetics with the wild-type strain. Furthermore, high viral loads were found in vivo regardless of the presence or absence of the ORF6 deletion. Our study highlights the transmission and replication capacity of two newly described deletion variants in the same ORF6 region. ImportanceWhile the SARS-CoV-2 genome has remained relatively stable since its emergence in the human population, genomic deletions are an evolutionary pattern previously described for the related SARS-CoV. Real-time genomic monitoring of the circulating variants is paramount to detect strain prevalence and transmission dynamics. Given the role of ORF6 in interferon modulation, further characterization, such as mechanistic interactions and interferon monitoring in patients, is crucial in understanding the viral-host factors driving disease evolution.

microbiology↗

bioRxiv2020

Evaluation of NGS-based approaches for SARS-CoV-2 whole genome characterisation

Since the beginning of the COVID-19 outbreak, SARS-CoV-2 whole-genome sequencing (WGS) has been performed at unprecedented rate worldwide with the use of very diverse Next Generation Sequencing (NGS) methods. Herein, we compare the performance of four NGS-based approaches for SARS-CoV-2 WGS. Twenty four clinical respiratory samples with a large scale of Ct values (from 10.7 to 33.9) were sequenced with four methods. Three used Illumina sequencing: an in-house metagenomic NGS (mNGS) protocol and two newly commercialized kits including a hybridization capture method developed by Illumina (DNA Prep with Enrichment kit and Respiratory Virus Oligo Panel, RVOP) and an amplicon sequencing method developed by Paragon Genomics (CleanPlex SARS-CoV-2 kit). We also evaluated the widely used amplicon sequencing protocol developed by ARTIC Network and combined with Oxford Nanopore Technologies (ONT) sequencing. All four methods yielded near-complete genomes (>99%) for high viral loads samples, with mNGS and RVOP producing the most complete genomes. For mid viral loads, 2/8 and 1/8 genomes were incomplete (<99%) with mNGS and both CleanPlex and RVOP, respectively. For low viral loads (Ct [≥]25), amplicon-based enrichment methods were the most sensitive techniques yielding complete genomes for 7/8 samples. All methods were highly concordant in terms of identity in complete consensus sequence. Just one mismatch in two samples was observed in CleanPlex vs the other methods, due to the dedicated bioinformatics pipeline setting a high threshold to call SNP compared to reference sequence. Importantly, all methods correctly identified a newly observed 34-nt deletion in ORF6 but required specific bioinformatic validation for RVOP. Finally, as a major warning for targeted techniques, a default of coverage in any given region of the genome should alert to a potential rearrangement or a SNP in primer annealing or probe-hybridizing regions and would require regular updates of the technique according to SARS-CoV-2 evolution.

genomics↗

bioRxiv2020

Cross talk between tumor stroma and cancer cells plays a critical role in progressive enrichment of cancer stem cell phenotype in primary breast tumors

In order to delineate the underlying molecular mechanisms responsible for intra tumoral enrichment of BCSCs in aggressive breast tumors, firstly we evaluated the frequency and characteristics of breast cancer stem cells (BCSCs) within the tumor mass as well as in pathologically normal adjacent tissues in primary breast carcinomas of various clinical and histological grades. Then, we evaluated the expression profiles of various genes in non-cancer stem cells from these tumors to delineate the role played by cellular niche in de novo origin and/or expansion of intra-tumoral cancer stem cells. The study included primary tumor and adjacent normal breast tissue specimens from chemotherapy-naive breast carcinoma patients. The BCSCs, identified as Lin-CD44+CD24- and aldehyde dehydrogenase 1 A1 positive were enumerated. The frequency of intra-tumoral BCSCs was correlated with various clinicopathological parameters of breast cancer. The flow-cytometrically sorted stromal cells and cancer cells from treatment naive primary breast tumors were processed for gene expression profiling using a custom designed PCR array of genes known to facilitate cancer cell proliferation and disease progression. The frequency of BCSCs within the tumor mass as well as in the adjacent normal tissue correlated significantly with histopathological and molecular grades of tumors indicating a direct relationship of BCSC with aggressive behavior of breast cancer. A significantly higher number of BCSCs was also detected in metastatic LN group as compared to non-metastatic LN. Further, a significantly increased expression of the genes associated with growth factors, cytokines & matricellular proteins in tumors with high BCSCs content (> 5%; Hi-BCSCs tumors) as compared to Lo-BCSC tumors (with <5% intratumoral BCSC content) suggested the possible contribution of stromal cells and cancer cells in intra-tumoral expansion of CSCs. Similarly, a significant up-regulation of genes associated with hypoxia and angiogenesis in Hi-BCSCs tumors further supported the role of hypoxic environment. The expression levels of genes associated with epithelial to mesenchymal transition also followed a similar pattern. On the other hand, downregulated SNAI1 gene (generally upregulated in onset of EMT) in stromal cells of Hi-BCSCs tumors suggests a post EMT environment in Hi-BCSCs tumors. The findings suggest that the molecular crosstalk between the non-BCSC niche cells and the cancer stem cells within the breast cancer microenvironment directly contribute to formation of biologically conducive conditions for expansion of cancer stem cells.

cancer biology↗

bioRxiv2020

Molecular characterization of SARS-CoV-2 in the first COVID-19 cluster in France reveals an amino-acid deletion in nsp2 (Asp268Del)

We present the first genetic characterization of a COVID-19 cluster in Europe using metagenomic next-generation sequencing (mNGS). Despite low viral loads, the mNGS workflow used herein allowed to characterize the whole genome sequences of SARS-CoV2 isolated from an asymptomatic patient, in 2 clinical samples collected 1 day apart. Comparison of these sequences suggests viral evolution with development of quasispecies. In addition, the present workflow identified a new deletion in nsp2 (Asp268Del) which was found in all 3 samples originating from this cluster as well as in 37 other viruses collected in England and in Netherlands, suggesting the spread of this deletion in Europe. The impact of Asp268Del on SARS-CoV-2 transmission and pathogenicity, as well as on PCR performances and anti-viral strategy should be rapidly evaluated in further studies.

microbiology↗

bioRxiv2020

SimpylCellCounter: An Automated Solution for Quantifying Cells in Brain Tissue

Rationale & ObjectiveManual quantification of activated cells can provide valuable information about stimuli-induced changes within brain regions; however, this analysis remains time intensive. Therefore, we created SimpylCellCounter (SCC), an automated method to quantify cells that express Cfos protein, an index of neuronal activity, in brain tissue and benchmarked it against two widely-used methods: OpenColonyFormingUnit (OCFU) and ImageJ Edge Detection Macro (IMJM). MethodsIn Experiment 1, manually-obtained counts were compared to those detected via OCFU, IMJM and SCC. The absolute error in counts (manual versus automated method) was calculated, and error types were categorized as false positives or negatives. In Experiment 2, performance analytics of OCFU, IMJM and SCC were compared. In Experiment 3, SCC performed analysis on images it was not trained on, to assess its general utility. Results & ConclusionsWe found SCC to be highly accurate and efficient in quantifying both cells with circular morphologies and those expressing Cfos. Additionally, SCC utilizes a new approach for counting overlapping cells with a pretrained convolutional neural network classifier. The current study demonstrates that SCC is a novel, automated tool to quantify cells in brain tissue, complementing current, open-sourced quantification methods designed to detect cells in vitro.

neuroscience↗

bioRxiv2019

Sustained exposure to trypsin causes cells to transition into a state of reversible stemness that is amenable to transdifferentiation

During cell culture, trypsin, a serine protease, is applied to cells for 5-10 minutes to separate them from each other and from the underlying substratum so that they can be transferred to a different vessel, for re-plating, after growth medium containing 10 % serum has been added to the cells, in a well-known technique known as passaging. The serum in the growth medium contains alpha-1 antitrypsin, which is a potent inhibitor of trypsin, elastase and other serine proteases. Although what is used is bovine serum in which levels of proteins could be different from levels seen in humans, normal human serum contains A1AT (> 1 mg/ml; > [~]18 {micro}mol/L) as well as trypsin itself (< 460 ng/ml, or [~]0.02 {micro}mol/L), with the former in a [~]900-fold molar excess over the latter. Thus, it may be assumed there is also enough A1AT in the bovine serum added during passaging, to neutralize the trypsin ([~]100 M) present in the small volume of trypsin-EDTA solution used to separate cells. What are the consequences of not adding serum, when growth medium is added, or of maintaining cells for a few tens of hours in the presence of trypsin, in a serum-free growth medium? What does such sustained exposure to trypsin during cell culture do to cells? More generally, what are the responses of cells within an organism to the balance of trypsin and A1AT in the serum that bathes them constantly? We know that excesses and deficiencies in the levels of either trypsin or A1AT are associated with disease. We know that cellular metabolism can be influenced through signaling involving protease activated membrane GPCR receptors (PAR1-4). In particular, we know of a receptor called PAR2, which is specifically activated by trypsin, expressed by cells at baseline levels, and upregulated through some feedback involving trypsin-activation. We also know that cells at sites of injury or inflammation produce and secrete trypsin, and that this trypsin can act locally upon cells in a variety of ways, all of which have probably not yet been elucidated. Here, we show that sustained exposure to trypsin induces cells to de-differentiate into a stem-like state. We show that if serum is either not added at all, or added and then washed away (after confluency is attained), during cell culture, all cells exposed to exogenously-added trypsin undergo changes in morphology, transcriptome, secretome, and developmental potential, and transition into a state of stemness, in minimal essential medium (MEM). Regardless of their origins, i.e., independent of whether they are derived from primary cultures, cell lines or cancer cell lines, and regardless of the original cell type used, exposure to trypsin ([~]10 {micro}M; [~]250 {micro}g/ml) at a concentration 10-fold lower than that used to separate cells during passaging ([~]100 M), over a period of 24-48 hours, causes cells to (1) become rounded, (2) cluster together, (3) get arrested in the G0/G1 stage of the cell cycle, (4) display increased presence of 5-hydroxymethyl cytosine in their nuclei (indicative of reprogramming), (5) display increased levels of activated PAR2 membrane receptor, (6) become capable of very efficient efflux of drug-mimicking dyes, (7) express factors and/or markers known to be associated with induction and/or attainment of stemness, with predominant expression of Sox-2 within cell nuclei; (8) display overall transcriptomic (RNASEQ) profiles characteristic of stemness; (9) secrete stemness-associated factors such as bFGF, and IL-1{beta}, into the medium, in quantities sufficient to support autocrine function (in certain cases); and (10) display increased conversion of pro-MMPs into activated MMPs in the cells secretome. Notably, (11) inclusion of differentiating and/or transdifferentiating factors in the environment of such cells causes them to express markers associated with ectodermal, endodermal and mesodermal cell lineages and/or transdifferentiate into specific cell types, e.g., adipocytes or osteocytes.\n\nMost intriguingly of all, (12) the attained stemness appears to be reversible, i.e., withdrawal of trypsin from the medium prior to addition of any differentiating factors restores cells to their original morphology, also over a period of 24-48 hours. Further, (13) a known PAR2 agonist, and a known PAR2 antagonist, respectively, appear to mimic effects of trypsin addition and withdrawal/inhibition. In addition, (14) in experiments with a particular cancer characterized by high levels of stemness (TNBC; triple negative breast cancer), tissues of all TNBC patients express high levels of the PAR2 receptor, as do cells from a known TNBC-derived cell line. We propose that through their effects on PAR levels, and PAR activation status, the balance of trypsin and A1AT levels in organisms normally regulates cellular potential for differentiation, de-differentiation or transdifferentiation, in a local manner, with the default status being that A1AT inhibits trypsin and keeps cells differentiated, whereas sustained trypsin signaling at sites of injury through local production of trypsin helps to place cells into an intermediate state of stemness from which they can either return to their original differentiated state(s), or undergo factor-dependent differentiation, or transdifferentiation, into specific cell types or lineages. It is also possible that reduction in A1AT promotes regeneration. We present a core (RNASEQ-derived) signature for trypsin-induced stemness in human corneal fibroblasts (HCFs) and cells from a retinal pigment epithelial cell line (ARPE-19), noting that there are commonalities as well as differences between them, which suggests that this core signature will be amended with RNASEQ studies of more trypsin-exposed cell types. Our findings offer a possible explanation for the recent unexplained increase in the preference for serum-free protocols used for induction and maintenance of stemness involving iPSCs and mesenchymal stem cells. Also, our studies suggest a new approach to understanding and exploiting how organisms might use stemness, in adults. Trypsin-dominated serine protease induced reprogramming (SPIR) might offer a more natural, and suitably softer, method of reprogramming of cellular developmental potential for local regenerative requirements in animal tissues.

cell biology↗

bioRxiv2018

Evolution of influenza genome diversity during infection in immunocompetent patients

IntroductionMinor frequency viruses play many important roles during viral infection that cannot be explained by the consensus sequence alone. In influenza, immunosuppressed individuals appear to generate numerous viral variants, leading to subpopulations with important role in infection. The objective of the present study was to describe viral diversification over time in immunocompetent patients during influenza virus infection.\n\nMethodsAll clinical records of patients admitted to the Lyon university hospital (Lyon, France) during the influenza infection epidemics of the 2010-2015 period and sampled at least twice during their clinical management were retrospectively analyzed. To estimate performance of the sequencing procedures, well-characterized plasmids containing each of the 8 segments of influenza viruses were used as quality controls. Diversity, i.e. the number of validated single nucleotide variants, was analyzed to compare characteristics over time and according to clinical severity (mild, severe with neurological complications, severe with respiratory complications).\n\nResultsAfter validation on quality controls (n=51), and verification of possible confusion bias, a 5%-threshold of detection was applied to clinical viral sequences (n=29). At this threshold, amino-acid coordinates (n=183/4,246, 4.31%) were identified as having at least one mutation during clinical course, independently of the clinical severity. Considering a threshold of 4 days of symptoms, as a limit for early and late sampling, diversity was significantly higher in late samples for the mild group, compared to both early mild and severe groups (p<0.05). At a single-segment scale, for PB2-coding segment, diversity was significantly higher in early samples of the neurological group than in both early and late samples in the respiratory group and for late samples in the mild group (p<0.05). For the NS1-coding segment, significant differences were observed between initial diversity of mild and severe patients, as for early and late samples in mild patients (p<0.01). Discussion. This study is the first describing diversity through time, associating biological and clinical information during viral diversification, during the infection of an immunocompetent human host. This latter opens a large field of investigation in infectious disease management using next-generation sequencing and suggest development of new therapies, focusing on non-antigenic viral properties, in non-vaccine fields of research

microbiology↗

bioRxiv2018

Quality control implementation for universal characterization of DNA and RNA viruses in clinical respiratory samples using single metagenomic next-generation sequencing workflow

BackgroundIn recent years, metagenomic Next-Generation Sequencing (mNGS) has increasingly been used for an accurate assumption-free virological diagnosis. However, the systematic workflow evaluation on clinical respiratory samples and implementation of quality controls (QCs) is still lacking.\n\nMethodsA total of 3 QCs were implemented and processed through the whole mNGS workflow: a notemplate-control to evaluate contamination issues during the process; an internal and an external QC to check the integrity of the reagents, equipment, the presence of inhibitors, and to allow the validation of results for each sample. The workflow was then evaluated on 37 clinical respiratory samples from patients with acute respiratory infections previously tested for a broad panel of viruses using semi-quantitative real-time PCR assays (28 positive samples including 6 multiple viral infections; 9 negative samples). Selected specimens included nasopharyngeal swabs (n = 20), aspirates (n = 10), or sputums (n = 7).\n\nResultsThe optimal spiking level of the internal QC was first determined in order to be sufficiently detected without overconsumption of sequencing reads. According to QC validation criteria, mNGS results were validated for 34/37 selected samples. For valid samples, viral genotypes were accurately determined for 36/36 viruses detected with PCR (viral genome coverage ranged from 0.6% to 100%, median = 67.7%). This mNGS workflow allowed the detection of DNA and RNA viruses up to a semi-quantitative PCR Ct value of 36. The six multiple viral infections involving 2 to 4 viruses were also fully characterized. A strong correlation between results of mNGS and real-time PCR was obtained for each type of viral genome (R2 ranged from 0.72 for linear single-stranded (ss) RNA viruses to 0.98 for linear ssDNA viruses).\n\nConclusionsAlthough the potential of mNGS technology is very promising, further evaluation studies are urgently needed for its routine clinical use within a reasonable timeframe. The approach described herein is crucial to bring standardization and to ensure the quality of the generated sequences in clinical setting. We provide an easy-to-use single protocol successfully evaluated for the characterization of a broad and representative panel of DNA and RNA respiratory viruses in various types of clinical samples.

microbiology↗