Introduction: CAR-T cell therapy is a novel approach in the treatment of hematological tumors. However, it is associated with life-threatening side effects, such as the severe cytokine release syndrome (sCRS). Therefore, predicting the occurrence and development of sCRS is of great significance for clinical CAR-T therapy. The study of existing clinical data by artificial intelligence may bring useful information.Methods: By analyzing the heat map of clinical factors and comparing them between severe and non-severe CRS, we can identify significant differences among these factors and understand their interrelationships. Ultimately, a decision tree approach was employed to predict the timing of severe CRS in both children and adults, considering variables such as the same day, the day before, and initial values.Results: We measured cytokines and clinical biomarkers in 202 patients who received CAR-T therapy. Peak levels of 25 clinical factors, including IFN-γ, IL6, IL10, ferritin, and D-dimer, were highly associated with severe CRS after CAR T cell infusion. Using the decision tree model, we were able to accurately predict which patients would develop severe CRS consisting of three clinical factors, classified as same-day, day-ahead, and initial value prediction. Changes in serum biomarkers, including C-reactive protein and ferritin, were associated with CRS, but did not alone predict the development of severe CRS.Conclusion: Our research will provide significant information for the timely prevention and treatment of sCRS, during CAR-T immunotherapy for tumors, which is essential to reduce the mortality rate of patients.

The ability of a transcription factor to regulate its targets is modulated by a variety of genetic and epigenetic mechanisms. Alternative splicing can modulate gene function by adding or removing certain protein domains, and therefore affect the activity of protein. Reverse engineering of gene regulatory networks using gene expression profiles has proven valuable in dissecting the logical relationships among multiple proteins during the transcriptional regulation. However, it is unclear whether alternative splicing of certain proteins affects the activity of other transcription factors. In order to investigate the roles of alternative splicing during transcriptional regulation, we constructed a statistical model to infer whether the alternative splicing events of modulator proteins can affect the ability of key transcription factors in regulating the expression levels of their transcriptional targets. We tested our strategy in KIRC (Kidney Renal Clear Cell Carcinoma) using the RNA-seq data downloaded from TCGA (the Cancer Genomic Atlas). We identified 828of modulation relationships between the splicing levels of modulator proteins and activity levels of transcription factors. For instance, we found that the activity levels of GR (glucocorticoid receptor) protein, a key transcription factor in kidney, can be influenced by the splicing status of multiple proteins, including TP53, MDM2 (mouse double minute 2 homolog), RBM14 (RNA-binding protein 14) and SLK (STE20 like kinase). The influenced GR-targets are enriched by key cancer-related pathways, including p53 signaling pathway, TR/RXR activation, CAR/RXR activation, G1/S checkpoint regulation pathway, and G2/M DNA damage checkpoint regulation pathway. Our analysis suggests, for the first time, that exon inclusion levels of certain regulatory proteins can affect the activities of many transcription factors. Such analysis can potentially unravel a novel mechanism of how splicing variation influences the cellular function and provide important insights for how dysregulation of splicing outcome can lead to various diseases.

While synonymous single-nucleotide variants (sSNVs) have largely been unstudied, since they do not alter protein sequence, mounting evidence suggests that they may affect RNA conformation, splicing, and the stability of nascent-mRNAs to promote various diseases. Accurately prioritizing deleterious sSNVs from a pool of neutral ones can significantly improve our ability of selecting functional genetic variants identified from various genome-sequencing projects, and, therefore, advance our understanding of disease etiology. In this study, we develop a computational algorithm to prioritize sSNVs based on their impact on mRNA splicing and protein function. In addition to genomic features that potentially affect splicing regulation, our proposed algorithm also includes dozens structural features that characterize the functions of alternatively spliced exons on protein function. Our systematical evaluation on thousands of sSNVs suggests that several structural features, including intrinsic disorder protein scores, solvent accessible surface areas, protein secondary structures, and known and predicted protein family domains, show significant differences between disease-causing and neutral sSNVs. Our result suggests that the protein structure features offer an added dimension of information while distinguishing disease-causing and neutral synonymous variants. The inclusion of structural features increases the predictive accuracy for functional sSNV prioritization.

The T cell receptor (TCR) repertoire is pivotal to the human immune system, and understanding its nuances can significantly enhance our ability to forecast cancer-related immune responses. However, existing methods often overlook the intra- and inter-sequence interactions of T cell receptors (TCRs), limiting the development of sequence-based cancer-related immune status predictions. To address this challenge, we propose BertTCR, an innovative deep learning framework designed to predict cancer-related immune status using TCRs. BertTCR combines a pre-trained protein large language model with deep learning architectures, enabling it to extract deeper contextual information from TCRs. Compared to three state-of-the-art sequence-based methods, BertTCR improves the AUC on an external validation set for thyroid cancer detection by 21 percentage points. Additionally, this model was trained on over 2000 publicly available TCR libraries covering 17 types of cancer and healthy samples, and it has been validated on multiple public external datasets for its ability to distinguish cancer patients from healthy individuals. Furthermore, BertTCR can accurately classify various cancer types and healthy individuals. Overall, BertTCR is the advancing method for cancer-related immune status forecasting based on TCRs, offering promising potential for a wide range of immune status prediction tasks.

With increasing trend of polypharmacy, drug‐drug interaction (DDI)‐induced adverse drug events (ADEs) are considered as a major challenge for clinical practice. As premarketing clinical trials usually have stringent inclusion/exclusion criteria, limited comedication data capture and often times small sample size have limited values in study DDIs. On the other hand, ADE reports collected by spontaneous reporting system (SRS) become an important source for DDI studies. There are two major challenges in detecting DDI signals from SRS: confounding bias and false positive rate. In this article, we propose a novel approach, propensity score‐adjusted three‐component mixture model (PS‐3CMM). This model can simultaneously adjust for confounding bias and estimate false discovery rate for all drug‐drug‐ADE combinations in FDA Adverse Event Reporting System (FAERS), which is a preeminent SRS database. In simulation studies, PS‐3CMM performs better in detecting true DDIs comparing to the existing approach. It is more sensitive in selecting the DDI signals that have nonpositive individual drug relative ADE risk (NPIRR). The application of PS‐3CMM is illustrated in analyzing the FAERS database. Compared to the existing approaches, PS‐3CMM prioritizes DDI signals differently. PS‐3CMM gives high priorities to DDI signals that have NPIRR. Both simulation studies and FAERS data analysis conclude that our new PS‐3CMM is a new method that is complement to the existing DDI signal detection methods.

Ion Torrent and Ion Proton are semiconductor-based sequencing technologies that feature rapid sequencing speed and low upfront and operating costs, thanks to the avoidance of modified nucleotides and optical measurements. Despite of these advantages, however, Ion semiconductor sequencing technologies suffer much reduced sequencing accuracy at the genomic loci with homopolymer repeats of the same nucleotide. Such limitation significantly reduces its efficiency for the biological applications aiming at accurately identifying various genetic variants. In this study, we propose a Bayesian inference-based method that takes the advantage of the signal distributions of the electrical voltages that are measured for all the homopolymers of a fixed length. By cross-referencing the length of homopolymers in the reference genome and the voltage signal distribution derived from the experiment, the proposed integrated model significantly improves the alignment accuracy around the homopolymer regions. Besides improving alignment accuracy on homopolymer regions for semiconductor-based sequencing technologies with the proposed model, similar strategies can also be used on other high-throughput sequencing technologies that share similar limitations.