Humanization of antibodies using a machine learning approach on large-scale repertoire data.
Ontology highlight
ABSTRACT: Motivation
Monoclonal antibody therapeutics are often produced from non-human sources (typically murine), and can therefore generate immunogenic responses in humans. Humanization procedures aim to produce antibody therapeutics that do not elicit an immune response and are safe for human use, without impacting efficacy. Humanization is normally carried out in a largely trial-and-error experimental process. We have built machine learning classifiers that can discriminate between human and non-human antibody variable domain sequences using the large amount of repertoire data now available.Results
Our classifiers consistently outperform the current best-in-class model for distinguishing human from murine sequences, and our output scores exhibit a negative relationship with the experimental immunogenicity of existing antibody therapeutics. We used our classifiers to develop a novel, computational humanization tool, Hu-mAb, that suggests mutations to an input sequence to reduce its immunogenicity. For a set of therapeutic antibodies with known precursor sequences, the mutations suggested by Hu-mAb show significant overlap with those deduced experimentally. Hu-mAb is therefore an effective replacement for trial-and-error humanization experiments, producing similar results in a fraction of the time.Availability
Hu-mAb (humanness scoring and humanization) is freely available to use at opig.stats.ox.ac.uk/webapps/humab.Supplementary information
Supplementary data are available at Bioinformatics online.
SUBMITTER: Marks C
PROVIDER: S-EPMC8760955 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Antibody humanization is a key step in the preclinical phase of the development of therapeutic antibodies, originally developed and tested in non-human models (most typically, in mouse). The standard technique of Complementarity-Determining Regions (CDR) grafting into human Framework Regions of germline sequences has some important drawbacks, in that the resulting sequences often need further back-mutations to ensure functionality and/or stability. Here we propose a new method to characterize the statistical distribution of the sequences of the variable regions of human antibodies, that takes into account phenotypical correlations between pairs of residues, both within and between chains. We define a "humanness score" of a sequence, comparing its performance in distinguishing human from murine sequences, with that of some alternative scores in the literature. We also compare the score with the experimental immunogenicity of clinically used antibodies. Finally, we use the humanness score as an optimization function and perform a search in the sequence space, starting from different murine sequences and keeping the CDR regions unchanged. Our results show that our humanness score outperforms other methods in sequence classification, and the optimization protocol is able to generate humanized sequences that are recognized as human by standard homology modelling tools.
| S-EPMC6172228 | biostudies-other
Project description:Techniques of data mining and machine learning were applied to a large database of medical and facility claims from commercially insured patients to determine the prevalence, gender demographics, and costs for individuals with provider-assigned diagnosis codes for myalgic encephalomyelitis (ME) or chronic fatigue syndrome (CFS). The frequency of diagnosis was 519-1,038/100,000 with the relative risk of females being diagnosed with ME or CFS compared to males 1.238 and 1.178, respectively. While the percentage of women diagnosed with ME/CFS is higher than the percentage of men, ME/CFS is not a "women's disease." Thirty-five to forty percent of diagnosed patients are men. Extrapolating from this frequency of diagnosis and based on the estimated 2017 population of the United States, a rough estimate for the number of patients who may be diagnosed with ME or CFS in the U.S. is 1.7 million to 3.38 million. Patients diagnosed with CFS appear to represent a more heterogeneous group than those diagnosed with ME. A machine learning model based on characteristics of individuals diagnosed with ME was developed and applied, resulting in a predicted prevalence of 857/100,000 (p > 0.01), or roughly 2.8 million in the U.S. Average annual costs for individuals with a diagnosis of ME or CFS were compared with those for lupus (all categories) and multiple sclerosis (MS), and found to be 50% higher for ME and CFS than for lupus or MS, and three to four times higher than for the general insured population. A separate aspect of the study attempted to determine if a diagnosis of ME or CFS could be predicted based on symptom codes in the insurance claims records. Due to the absence of specific codes for some core symptoms, we were unable to validate that the information in insurance claims records is sufficient to identify diagnosed patients or suggest that a diagnosis of ME or CFS should be considered based solely on looking for presence of those symptoms. These results show that a prevalence rate of 857/100,000 for ME/CFS is not unreasonable; therefore, it is not a rare disease, but in fact a relatively common one.
| S-EPMC6331450 | biostudies-other
Project description:Nationwide population-based cohort provides a new opportunity to build an automated risk prediction model based on individuals' history of health and healthcare beyond existing risk prediction models. We tested the possibility of machine learning models to predict future incidence of Alzheimer's disease (AD) using large-scale administrative health data. From the Korean National Health Insurance Service database between 2002 and 2010, we obtained de-identified health data in elders above 65 years (N = 40,736) containing 4,894 unique clinical features including ICD-10 codes, medication codes, laboratory values, history of personal and family illness and socio-demographics. To define incident AD we considered two operational definitions: "definite AD" with diagnostic codes and dementia medication (n = 614) and "probable AD" with only diagnosis (n = 2026). We trained and validated random forest, support vector machine and logistic regression to predict incident AD in 1, 2, 3, and 4 subsequent years. For predicting future incidence of AD in balanced samples (bootstrapping), the machine learning models showed reasonable performance in 1-year prediction with AUC of 0.775 and 0.759, based on "definite AD" and "probable AD" outcomes, respectively; in 2-year, 0.730 and 0.693; in 3-year, 0.677 and 0.644; in 4-year, 0.725 and 0.683. The results were similar when the entire (unbalanced) samples were used. Important clinical features selected in logistic regression included hemoglobin level, age and urine protein level. This study may shed a light on the utility of the data-driven machine learning model based on large-scale administrative health data in AD risk prediction, which may enable better selection of individuals at risk for AD in clinical trials or early detection in clinical settings.
| S-EPMC7099065 | biostudies-literature
Project description:BackgroundMetagenomics is an approach to the characterization of microbial genomes via the direct isolation of genomic sequences from the environment without prior cultivation. The amount of metagenomic sequence data is growing fast while computational methods for metagenome analysis are still in their infancy. In contrast to genomic sequences of single species, which can usually be assembled and analyzed by many available methods, a large proportion of metagenome data remains as unassembled anonymous sequencing reads. One of the aims of all metagenomic sequencing projects is the identification of novel genes. Short length, for example, Sanger sequencing yields on average 700 bp fragments, and unknown phylogenetic origin of most fragments require approaches to gene prediction that are different from the currently available methods for genomes of single species. In particular, the large size of metagenomic samples requires fast and accurate methods with small numbers of false positive predictions.ResultsWe introduce a novel gene prediction algorithm for metagenomic fragments based on a two-stage machine learning approach. In the first stage, we use linear discriminants for monocodon usage, dicodon usage and translation initiation sites to extract features from DNA sequences. In the second stage, an artificial neural network combines these features with open reading frame length and fragment GC-content to compute the probability that this open reading frame encodes a protein. This probability is used for the classification and scoring of gene candidates. With large scale training, our method provides fast single fragment predictions with good sensitivity and specificity on artificially fragmented genomic DNA. Additionally, this method is able to predict translation initiation sites accurately and distinguishes complete from incomplete genes with high reliability.ConclusionLarge scale machine learning methods are well-suited for gene prediction in metagenomic DNA fragments. In particular, the combination of linear discriminants and neural networks is promising and should be considered for integration into metagenomic analysis pipelines. The data sets can be downloaded from the URL provided (see Availability and requirements section).
| S-EPMC2409338 | biostudies-literature
Project description:The goal of this work is to introduce new metrics to assess risk of Alzheimer's disease (AD) which we call AD Pattern Similarity (AD-PS) scores. These metrics are the conditional probabilities modeled by large-scale regularized logistic regression. The AD-PS scores derived from structural MRI and cognitive test data were tested across different situations using data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study. The scores were computed across groups of participants stratified by cognitive status, age and functional status. Cox proportional hazards regression was used to evaluate associations with the distribution of conversion times from mild cognitive impairment to AD. The performances of classifiers developed using data from different types of brain tissue were systematically characterized across cognitive status groups. We also explored the performance of anatomical and cognitive-anatomical composite scores generated by combining the outputs of classifiers developed using different types of data. In addition, we provide the AD-PS scores performance relative to other metrics used in the field including the Spatial Pattern of Abnormalities for Recognition of Early AD (SPARE-AD) index and total hippocampal volume for the variables examined.
| S-EPMC3826736 | biostudies-literature
Project description:BackgroundThe field of psychiatry has long sought biomarkers that can objectively diagnose patients, predict treatment response, or identify individuals at risk of illness onset. However, reliable psychiatric biomarkers have yet to emerge. The recent application of machine learning techniques to develop neuroimaging-based biomarkers has yielded promising preliminary results. However, much of the work in this domain has not met best practice standards from the field of machine learning. This is especially true for studies of anxiety, creating uncertainty about the potential for anxiety biomarker development.MethodsWe applied machine learning tools to predict trait anxiety from neuroimaging measurements in humans. Using publicly available data from the Brain Genomics Superstruct Project, we compared a suite of neuroimaging-based machine learning models predicting anxiety within a discovery sample (n = 531, 307 women) via k-fold cross-validation, and we tested the final model (a stacked model incorporating region-to-region functional connectivity, amygdala seed-to-voxel connectivity, and volumetric and cortical thickness data) in a held-out, unseen test sample (n = 348, 209 women).ResultsThough the best model was able to predict anxiety within the discovery sample (cross-validated R2 of .06, permutation test p < .001), the generalization test within the holdout sample failed (R2 of -.04, permutation test p > .05).ConclusionsIn this study, we did not find evidence of a generalizable anxiety biomarker. However, we encourage other researchers to investigate this topic, utilizing large samples and proper methodology, to clarify the potential of neuroimaging-based anxiety biomarkers.
| S-EPMC6925354 | biostudies-literature
Project description:Freely-available satellite data streams and the ability to process these data on cloud-computing platforms such as Google Earth Engine have made frequent, large-scale landcover mapping at high resolution a real possibility. In this paper we apply these technologies, along with machine learning, to the mapping of peatlands-a landcover class that is critical for preserving biodiversity, helping to address climate change impacts, and providing ecosystem services, e.g., carbon storage-in the Boreal Forest Natural Region of Alberta, Canada. We outline a data-driven, scientific framework that: compiles large amounts of Earth observation data sets (radar, optical, and LiDAR); examines the extracted variables for suitability in peatland modelling; optimizes model parameterization; and finally, predicts peatland occurrence across a large boreal area (397, 958 km2) of Alberta at 10 m spatial resolution (equalling 3.9 billion pixels across Alberta). The resulting peatland occurrence model shows an accuracy of 87% and a kappa statistic of 0.57 when compared to our validation data set. Differentiating peatlands from mineral wetlands achieved an accuracy of 69% and kappa statistic of 0.37. This data-driven approach is applicable at large geopolitical scales (e.g., provincial, national) for wetland and landcover inventories that support long-term, responsible resource management.
| S-EPMC6576777 | biostudies-literature
Project description:A large number of highly pathogenic bacteria utilize secretion systems to translocate effector proteins into host cells. Using these effectors, the bacteria subvert host cell processes during infection. Legionella pneumophila translocates effectors via the Icm/Dot type-IV secretion system and to date, approximately 100 effectors have been identified by various experimental and computational techniques. Effector identification is a critical first step towards the understanding of the pathogenesis system in L. pneumophila as well as in other bacterial pathogens. Here, we formulate the task of effector identification as a classification problem: each L. pneumophila open reading frame (ORF) was classified as either effector or not. We computationally defined a set of features that best distinguish effectors from non-effectors. These features cover a wide range of characteristics including taxonomical dispersion, regulatory data, genomic organization, similarity to eukaryotic proteomes and more. Machine learning algorithms utilizing these features were then applied to classify all the ORFs within the L. pneumophila genome. Using this approach we were able to predict and experimentally validate 40 new effectors, reaching a success rate of above 90%. Increasing the number of validated effectors to around 140, we were able to gain novel insights into their characteristics. Effectors were found to have low G+C content, supporting the hypothesis that a large number of effectors originate via horizontal gene transfer, probably from their protozoan host. In addition, effectors were found to cluster in specific genomic regions. Finally, we were able to provide a novel description of the C-terminal translocation signal required for effector translocation by the Icm/Dot secretion system. To conclude, we have discovered 40 novel L. pneumophila effectors, predicted over a hundred additional highly probable effectors, and shown the applicability of machine learning algorithms for the identification and characterization of bacterial pathogenesis determinants.
| S-EPMC2701608 | biostudies-literature
Project description:BackgroundIn recent years the field of movement ecology has been revolutionized by our ability to collect high-accuracy, fine scale telemetry data from individual animals and groups. This growth in our data collection capacity has led to the development of statistical techniques that integrate telemetry data with random walk models to infer key parameters of the movement dynamics. While much progress has been made in the use of these models, several challenges remain. Notably robust and scalable methods are required for quantifying parameter uncertainty, coping with intermittent location fixes, and analysing the very large volumes of data being generated.MethodsIn this work we implement a novel approach to movement modelling through the use of multilevel Gaussian processes. The hierarchical structure of the method enables the inference of continuous latent behavioural states underlying movement processes. For efficient inference on large data sets, we approximate the full likelihood using trajectory segmentation and sample from posterior distributions using gradient-based Markov chain Monte Carlo methods.ResultsWhile formally equivalent to many continuous-time movement models, our Gaussian process approach provides flexible, powerful models that can detect multiscale patterns and trends in movement trajectory data. We illustrate a further advantage to our approach in that inference can be performed using highly efficient, GPU-accelerated machine learning libraries.ConclusionsMultilevel Gaussian process models offer efficient inference for large-volume movement data sets, along with the fitting of complex flexible models. Applications of this approach include inferring the mean location of a migration route and quantifying significant changes, detecting diurnal activity patterns, or identifying the onset of directed persistent movements.
| S-EPMC7893961 | biostudies-literature
Project description:Using data from 183 public human data sets from PRIDE, a machine learning model was trained to identify tissue and cell-type specific protein patterns. PRIDE projects were searched with ionbot and tissue/cell type annotation was manually added. Data from physiological samples were used to train a Random Forest model on protein abundances to classify samples into tissues and cell types. Subsequently, a one-vs-all classification and feature importance were used to analyze the most discriminating protein abundances per class. Based on protein abundance alone, the model was able to predict tissues with 98% accuracy, and cell types with 99% accuracy. The F-scores describe a clear view on tissue-specific proteins and tissue-specific protein expression patterns. In-depth feature analysis shows slight confusion between physiologically similar tissues, demonstrating the capacity of the algorithm to detect biologically relevant patterns. These results can in turn inform downstream uses, from identification of the tissue of origin of proteins in complex samples such as liquid biopsies, to studying the proteome of tissue-like samples such as organoids and cell lines.
| S-EPMC10088018 | biostudies-literature