Showing papers by "Transgene SA published in 2021"
TL;DR: BERTran et al. as mentioned in this paper presented a transformer neural network model which leverages self-supervised pretraining from a large corpus of protein sequences, and proposed a multiple instance learning (MIL) framework to deconvolve mass spectrometry data where multiple potential MHC alleles may have presented each peptide.
Abstract: Motivation Increasingly comprehensive characterisation of cancer-associated genetic alterations has paved the way for the development of highly specific therapeutic vaccines. Predicting precisely the binding and presentation of peptides to MHC alleles is an important step towards such therapies. Recent data suggest that presentation of both class I and II epitopes are critical for the induction of a sustained effective immune response. However, the prediction performance for MHC class II has been limited compared to class I. Results We present a transformer neural network model which leverages self-supervised pretraining from a large corpus of protein sequences. We also propose a multiple instance learning (MIL) framework to deconvolve mass spectrometry data where multiple potential MHC alleles may have presented each peptide. We show that pretraining boosted the performance for these tasks. Combining pretraining and the novel MIL approach, our model outperforms state-of-the-art models based on peptide and MHC sequence only for both binding and cell surface presentation predictions. Availability Our source code is available at https://github.com/s6juncheng/BERTMHC under a non-commercial license. A webserver is available at https://bertmhc.privacy.nlehd.de/.
17 citations
TL;DR: In this paper, the authors evaluated the correlation of HPV16 DNA, RNA transcripts and features of adaptive immune response by evaluating antibody isotypes against E2, E7 antigens and density of tumor-infiltrating lymphocytes (TIL).
Abstract: Introduction Human papillomavirus 16 (HPV16) is the main cause of oropharyngeal squamous cell carcinoma (OPSCC). To date, the links between HPV16 gene expression and adaptive immune responses have not been investigated. We evaluated the correlation of HPV16 DNA, RNA transcripts and features of adaptive immune response by evaluating antibody isotypes against E2, E7 antigens and density of tumor-infiltrating lymphocytes (TIL). Material and methods FFPE-tissue from 27/77 p16-positive OPSCC patients was available. DNA and RNA were extracted and quantified using qPCR for all HPV16 genes. The TIL status was assessed. Immune responses against E2 and E7 were quantified by ELISA (IgG, IgA, and IgM; 77 serum samples pre-treatment, 36 matched post-treatment). Results Amounts of HPV16 genes were highly correlated at DNA and RNA levels. RNA co-expression of all genes was detected in 37% (7/19). E7 qPCR results were correlated with higher anti-E7 antibody (IgG, IgA) level in the blood. Patients with high anti-E2 IgG antibody (>median) had better overall survival (p=0.0311); anti-E2 and anti-E7 IgA levels had no detectable effect. During the first 6 months after treatment, IgA but not IgG increased significantly, and >6 months both antibody classes declined over time. Patients with immune cell-rich tumors had higher levels of circulating antibodies against HPV antigens. Conclusion We describe an HPV16 qPCR assay to quantify genomic and transcriptomic expression and correlate this with serum antibody levels against HPV16 oncoproteins. Understanding DNA/RNA expression, relationship to the antibody response in patients regarding treatment and outcome offers an attractive tool to improve patient care.
1 citations