Improving AlphaFold modeling using implicit information from experimental density maps

Machine learning prediction algorithms such as AlphaFold can create remarkably accurate protein models, but these models usually have some regions that are predicted with low confidence or poor accuracy. We hypothesized that by implicitly including experimental information, a greater portion of a model could be predicted accurately, and that this might synergistically improve parts of the model that were not fully addressed by either machine learning or experiment alone. An iterative procedure was developed in which AlphaFold models are automatically rebuilt based on experimental density maps and the rebuilt models are used as templates in new AlphaFold predictions. We find that including experimental information improves prediction beyond the improvement obtained with simple rebuilding guided by the experimental data. This procedure for AlphaFold modeling with density has been incorporated into an automated procedure for crystallographic and electron cryo-microscopy map interpretation.

Privacy concerns with using public data for suicide risk prediction algorithms: a public opinion survey of contextual appropriateness

Purpose Existing algorithms for predicting suicide risk rely solely on data from electronic health records, but such models could be improved through the incorporation of publicly available socioeconomic data – such as financial, legal, life event and sociodemographic data. The purpose of this study is to understand the complex ethical and privacy implications of incorporating sociodemographic data within the health context. This paper presents results from a survey exploring what the general public’s knowledge and concerns are about such publicly available data and the appropriateness of using it in suicide risk prediction algorithms. Design/methodology/approach A survey was developed to measure public opinion about privacy concerns with using socioeconomic data across different contexts. This paper presented respondents with multiple vignettes that described scenarios situated in medical, private business and social media contexts, and asked participants to rate their level of concern over the context and what factor contributed most to their level of concern. Specific to suicide prediction, this paper presented respondents with various data attributes that could potentially be used in the context of a suicide risk algorithm and asked participants to rate how concerned they would be if each attribute was used for this purpose. Findings The authors found considerable concern across the various contexts represented in their vignettes, with greatest concern in vignettes that focused on the use of personal information within the medical context. Specific to the question of incorporating socioeconomic data within suicide risk prediction models, the results of this study show a clear concern from all participants in data attributes related to income, crime and court records, and assets. Data about one’s household were also particularly concerns for the respondents, suggesting that even if one might be comfortable with their own being used for risk modeling, data about other household members is more problematic. Originality/value Previous studies on the privacy concerns that arise when integrating data pertaining to various contexts of people’s lives into algorithmic and related computational models have approached these questions from individual contexts. This study differs in that it captured the variation in privacy concerns across multiple contexts. Also, this study specifically assessed the ethical concerns related to a suicide prediction model and determining people’s awareness of the publicness of select data attributes, as well as which of these data attributes generated the most concern in such a context. To the best of the authors’ knowledge, this is the first study to pursue this question.

Prediction of frictional braking noise based on brake dynamometer test and artificial intelligent algorithms

Based on brake noise dynamometer test data, combined with the artificial intelligent algorithms, frictional braking noise is quantitatively analyzed and predicted in this study. To achieve this goal, a frictional braking noise prediction method is indicatively proposed, which consists of two main parts: first, based on the experimental data obtained from the brake noise dynamometer tests, and combining with the improved Long-Short-Term Memory (LSTM) algorithm, the coefficients of friction (COFs) are predicted under various braking test conditions. Then, based on the predicted braking COFs and other selected critical braking parameters, the quantitative prediction of frictional braking noise is obtained by means of the optimized eXtreme Gradient Boosting (XGBoost) algorithm. Finally, the inherent features of the XGBoost algorithm are employed to qualitatively analyze the importance of the main factors affecting the frictional braking noise. The prediction algorithms of COFs and frictional braking noise are validated by the brake dynamomter test data, and the R2 (R square) scores of both the LSTM and XGBoost prediction algorithms are 0.9, which verifies the feasibility of both algorithms. The main contribution of this work is to predict the braking noise based on a large set of test data and combined with the LSTM and XGBoost artificial intelligent algorithms, which can significantly save time for the brake system development and braking performance testing, and has significance to the rapid prediction of braking frictional noise and fast NVH (noise, vibration, and harshness) optimal design of frictional braking systems.

piCRISPR: Physically Informed Features Improve Deep Learning Models for CRISPR/Cas9 Off-Target Cleavage Prediction

CRISPR/Cas programmable nuclease systems have become ubiquitous in the field of gene editing. With progressing development, applications in in vivo therapeutic gene editing are increasingly within reach, yet limited by possible adverse side effects from unwanted edits. Recent years have thus seen continuous development of off-target prediction algorithms trained on in vitro cleavage assay data gained from immortalised cell lines. Here, we implement novel deep learning algorithms and feature encodings for off-target prediction and systematically sample the resulting model space in order to find optimal models and inform future modelling efforts. We lay emphasis on physically informed features, hence terming our approach piCRISPR, which we gain on the large, diverse crisprSQL off-target cleavage dataset. We find that our best-performing model highlights the importance of sequence context and chromatin accessibility for cleavage prediction and outperforms state-of-the-art prediction algorithms in terms of area under precision-recall curve.

IMMU-31. TUMOR-SPECIFIC ALTERNATIVE SPLICING GENERATES SPATIALLY-CONSERVED HLA-BINDING NEOANTIGEN TARGETS DETECTED THROUGH INTEGRATIVE TRANSCRIPTOMIC AND PROTEOMIC ANALYSES

BACKGROUND While immunotherapy is profoundly efficacious in certain cancers, its success is limited in cancers with lower mutational burden, such as gliomas. Therefore, investigating neoantigens beyond those from somatic mutations can expand the repertoire of immunotherapy targets. Recent studies detected alternative-splicing (AS) events in various cancer types that could potentially translate into tumor-specific proteins. Our study investigates AS within glioma to identify novel MHC-I-presented neoantigen targets through an integrative transcriptomic and proteomic computational pipeline, complemented by an extensive spatiotemporal analysis of the AS candidates. METHODS Bulk RNA-seq of high tumor purity TCGA-GBM/LGG (n=429) were analyzed through a novel systematic pipeline, and tumor-specific splicing junctions (neojunctions) were identified in silico by cross-referencing with bulk RNA-seq of GTEx normal tissue (n=9,166). Two HLA-binding prediction algorithms were subsequently incorporated to predict peptide sequences with high likelihood for HLA-presentation. Investigation of the tumor-wide clonality and temporal stability of the candidates was performed on extensive RNA-seq data from our spatially mapped intratumoral samples and longitudinally collected tumor tissue RNA-seq. Proteomic validation was conducted through mass-spec analysis of the Clinical Proteomic Tumor Analysis Consortium (CPTAC)-GBM repository (n=99). RESULTS Our analysis of TCGA-GBM/LGG bulk RNA-seq identified 249 putative neojunctions that translate into 222 cancer-specific peptide sequences which confer 21,489 tumor-specific n-mers (8-11 amino acids in length). Both prediction algorithms concurrently identified 271 n-mers likely to bind and be presented by HLA*A0101, HLA*A0201, HLA*A0301, HLA*A1101, or HLA*A2402. We confirmed the expression of 15 out of 58 HLA*A0201-binding candidates in HLA*A0201+ patient-derived glioma cell line RNA-seq with a subset of candidates conserved spatially. Analysis of CPTAC-GBM mass-spec data detected 23 tumor-specific peptides with 5 containing detected n-mers highly predicted to be HLA-presented. CONCLUSION Tumor-specific neojunctions identified in our unique integrative pipeline present novel candidate immunotherapy targets for gliomas and offer a new avenue in neoantigen discovery across cancer types.