SimAlign: High Quality Word Alignments Without Parallel Training Data Using Static and Contextualized Embeddings

BackgroundAccurately identified neoantigens can be effective therapeutic agents in both adjuvant and neoadjuvant settings. A key challenge for neoantigen discovery has been the availability of accurate prediction models for MHC peptide presentation. We have shown previously that our proprietary model based on (i) large-scale, in-house mono-allelic data, (ii) custom features that model antigen processing, and (iii) advanced machine learning algorithms has strong performance. We have extended upon our work by systematically integrating large quantities of high-quality, publicly available data, implementing new modelling algorithms, and rigorously testing our models. These extensions lead to substantial improvements in performance and generalizability. Our algorithm, named Systematic HLA Epitope Ranking Pan Algorithm (SHERPA™), is integrated into the ImmunoID NeXT Platform®, our immuno-genomics and transcriptomics platform specifically designed to enable the development of immunotherapies.MethodsIn-house immunopeptidomic data was generated using stably transfected HLA-null K562 cells lines that express a single HLA allele of interest, followed by immunoprecipitation using W6/32 antibody and LC-MS/MS. Public immunopeptidomics data was downloaded from repositories such as MassIVE and processed uniformly using in-house pipelines to generate peptide lists filtered at 1% false discovery rate. Other metrics (features) were either extracted from source data or generated internally by re-processing samples utilizing the ImmunoID NeXT Platform.ResultsWe have generated large-scale and high-quality immunopeptidomics data by using approximately 60 mono-allelic cell lines that unambiguously assign peptides to their presenting alleles to create our primary models. Briefly, our primary ‘binding’ algorithm models MHC-peptide binding using peptide and binding pockets while our primary ‘presentation’ model uses additional features to model antigen processing and presentation. Both primary models have significantly higher precision across all recall values in multiple test data sets, including mono-allelic cell lines and multi-allelic tissue samples. To further improve the performance of our model, we expanded the diversity of our training set using high-quality, publicly available mono-allelic immunopeptidomics data. Furthermore, multi-allelic data was integrated by resolving peptide-to-allele mappings using our primary models. We then trained a new model using the expanded training data and a new composite machine learning architecture. The resulting secondary model further improves performance and generalizability across several tissue samples.ConclusionsImproving technologies for neoantigen discovery is critical for many therapeutic applications, including personalized neoantigen vaccines, and neoantigen-based biomarkers for immunotherapies. Our new and improved algorithm (SHERPA) has significantly higher performance compared to a state-of-the-art public algorithm and furthers this objective.

Download Full-text

Optimal Scale of Hierarchical Image Segmentation with Scribbles Guidance for Weakly Supervised Semantic Segmentation

International Journal of Pattern Recognition and Artificial Intelligence ◽

10.1142/s0218001421540264 ◽

2021 ◽

pp. 2154026

Author(s):

Zaid Al-Huda ◽

Donghai Zhai ◽

Yan Yang ◽

Riyadh Nazar Ali Algburi

Keyword(s):

Image Segmentation ◽

Graphical Model ◽

Semantic Segmentation ◽

Saliency Map ◽

Training Data ◽

Deep Convolutional Neural Networks ◽

High Quality ◽

Optimal Scale ◽

Supervised Segmentation ◽

Weakly Supervised

Deep convolutional neural networks (DCNNs) trained on the pixel-level annotated images have achieved improvements in semantic segmentation. Due to the high cost of labeling training data, their applications may have great limitation. However, weakly supervised segmentation approaches can significantly reduce human labeling efforts. In this paper, we introduce a new framework to generate high-quality initial pixel-level annotations. By using a hierarchical image segmentation algorithm to predict the boundary map, we select the optimal scale of high-quality hierarchies. In the initialization step, scribble annotations and the saliency map are combined to construct a graphic model over the optimal scale segmentation. By solving the minimal cut problem, it can spread information from scribbles to unmarked regions. In the training process, the segmentation network is trained by using the initial pixel-level annotations. To iteratively optimize the segmentation, we use a graphical model to refine segmentation masks and retrain the segmentation network to get more precise pixel-level annotations. The experimental results on Pascal VOC 2012 dataset demonstrate that the proposed framework outperforms most of weakly supervised semantic segmentation methods and achieves the state-of-the-art performance, which is [Formula: see text] mIoU.

Download Full-text

Using diffusion MRI data acquired with ultra-high gradients to improve tractography in routine-quality data

10.1101/2021.06.28.450265 ◽

2021 ◽

Author(s):

Chiara Maffei ◽

Christine Lee ◽

Michael Planich ◽

Manisha Ramprasad ◽

Nivedita Ravi ◽

...

Keyword(s):

Diffusion Mri ◽

Region Of Interest ◽

Training Data ◽

Quality Data ◽

High Quality ◽

Scanning Time ◽

Human Connectome Project ◽

High Gradients ◽

Comprehensive Comparison

The development of scanners with ultra-high gradients, spearheaded by the Human Connectome Project, has led to dramatic improvements in the spatial, angular, and diffusion resolution that is feasible for in vivo diffusion MRI acquisitions. The improved quality of the data can be exploited to achieve higher accuracy in the inference of both microstructural and macrostructural anatomy. However, such high-quality data can only be acquired on a handful of Connectom MRI scanners worldwide, while remaining prohibitive in clinical settings because of the constraints imposed by hardware and scanning time. In this study, we first update the classical protocols for tractography-based, manual annotation of major white-matter pathways, to adapt them to the much greater volume and variability of the streamlines that can be produced from today's state-of-the-art diffusion MRI data. We then use these protocols to annotate 42 major pathways manually in data from a Connectom scanner. Finally, we show that, when we use these manually annotated pathways as training data for global probabilistic tractography with anatomical neighborhood priors, we can perform highly accurate, automated reconstruction of the same pathways in much lower-quality, more widely available diffusion MRI data. The outcomes of this work include both a new, comprehensive atlas of WM pathways from Connectom data, and an updated version of our tractography toolbox, TRActs Constrained by UnderLying Anatomy (TRACULA), which is trained on data from this atlas. Both the atlas and TRACULA are distributed publicly as part of FreeSurfer. We present the first comprehensive comparison of TRACULA to the more conventional, multi-region-of-interest approach to automated tractography, and the first demonstration of training TRACULA on high-quality, Connectom data to benefit studies that use more modest acquisition protocols.

Download Full-text

AI-Sketcher : A Deep Generative Model for Producing High-Quality Sketches

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v33i01.33012564 ◽

2019 ◽

Vol 33 ◽

pp. 2564-2571 ◽

Cited By ~ 3

Author(s):

Nan Cao ◽

Xin Yan ◽

Yang Shi ◽

Chaoran Chen

Keyword(s):

Artificial Intelligence ◽

Large Scale ◽

State Of The Art ◽

Positional Information ◽

Generative Model ◽

Training Data ◽

Creative Design ◽

High Quality ◽

User Input ◽

Variational Autoencoder

Sketch drawings play an important role in assisting humans in communication and creative design since ancient period. This situation has motivated the development of artificial intelligence (AI) techniques for automatically generating sketches based on user input. Sketch-RNN, a sequence-to-sequence variational autoencoder (VAE) model, was developed for this purpose and known as a state-of-the-art technique. However, it suffers from limitations, including the generation of lowquality results and its incapability to support multi-class generations. To address these issues, we introduced AI-Sketcher, a deep generative model for generating high-quality multiclass sketches. Our model improves drawing quality by employing a CNN-based autoencoder to capture the positional information of each stroke at the pixel level. It also introduces an influence layer to more precisely guide the generation of each stroke by directly referring to the training data. To support multi-class sketch generation, we provided a conditional vector that can help differentiate sketches under various classes. The proposed technique was evaluated based on two large-scale sketch datasets, and results demonstrated its power in generating high-quality sketches.

Download Full-text

Improving the Quality of Positive Datasets for the Establishment of Machine Learning Models for pre-microRNA Detection

Journal of Integrative Bioinformatics ◽

10.1515/jib-2017-0032 ◽

2017 ◽

Vol 14 (2) ◽

Author(s):

Müşerref Duygu Saçar Demirci ◽

Jens Allmer

Keyword(s):

Machine Learning ◽

Training Data ◽

Quality Data ◽

Virus Infections ◽

High Quality ◽

Disease States ◽

Positive Data ◽

Mirna Detection ◽

Post Transcriptional Regulation

AbstractMicroRNAs (miRNAs) are involved in the post-transcriptional regulation of protein abundance and thus have a great impact on the resulting phenotype. It is, therefore, no wonder that they have been implicated in many diseases ranging from virus infections to cancer. This impact on the phenotype leads to a great interest in establishing the miRNAs of an organism. Experimental methods are complicated which led to the development of computational methods for pre-miRNA detection. Such methods generally employ machine learning to establish models for the discrimination between miRNAs and other sequences. Positive training data for model establishment, for the most part, stems from miRBase, the miRNA registry. The quality of the entries in miRBase has been questioned, though. This unknown quality led to the development of filtering strategies in attempts to produce high quality positive datasets which can lead to a scarcity of positive data. To analyze the quality of filtered data we developed a machine learning model and found it is well able to establish data quality based on intrinsic measures. Additionally, we analyzed which features describing pre-miRNAs could discriminate between low and high quality data. Both models are applicable to data from miRBase and can be used for establishing high quality positive data. This will facilitate the development of better miRNA detection tools which will make the prediction of miRNAs in disease states more accurate. Finally, we applied both models to all miRBase data and provide the list of high quality hairpins.

Download Full-text

A Sequence-Based Novel Approach for Quality Evaluation of Third-Generation Sequencing Reads

Genes ◽

10.3390/genes10010044 ◽

2019 ◽

Vol 10 (1) ◽

pp. 44 ◽

Cited By ~ 1

Author(s):

Wenjing Zhang ◽

Neng Huang ◽

Jiantao Zheng ◽

Xingyu Liao ◽

Jianxin Wang ◽

...

Keyword(s):

Quality Evaluation ◽

Training Data ◽

Third Generation ◽

Contig Assembly ◽

High Quality ◽

Promising Alternative ◽

Third Generation Sequencing ◽

Long Reads ◽

Generation Sequencing

The advent of third-generation sequencing (TGS) technologies, such as the Pacific Biosciences (PacBio) and Oxford Nanopore machines, provides new possibilities for contig assembly, scaffolding, and high-performance computing in bioinformatics due to its long reads. However, the high error rate and poor quality of TGS reads provide new challenges for accurate genome assembly and long-read alignment. Efficient processing methods are in need to prioritize high-quality reads for improving the results of error correction and assembly. In this study, we proposed a novel Read Quality Evaluation and Selection Tool (REQUEST) for evaluating the quality of third-generation long reads. REQUEST generates training data of high-quality and low-quality reads which are characterized by their nucleotide combinations. A linear regression model was built to score the quality of reads. The method was tested on three datasets of different species. The results showed that the top-scored reads prioritized by REQUEST achieved higher alignment accuracies. The contig assembly results based on the top-scored reads also outperformed conventional approaches that use all reads. REQUEST is able to distinguish high-quality reads from low-quality ones without using reference genomes, making it a promising alternative sequence-quality evaluation method to alignment-based algorithms.

Download Full-text

A Novel Contrast Co-learning Framework for Generating High Quality Training Data

2010 IEEE International Conference on Data Mining ◽

10.1109/icdm.2010.23 ◽

2010 ◽

Author(s):

Zeyu Zheng ◽

Jun Yan ◽

Shuicheng Yan ◽

Ning Liu ◽

Zheng Chen ◽

...

Keyword(s):

Training Data ◽

High Quality ◽

Learning Framework

Download Full-text

Facing small and biased data dilemma in drug discovery with federated learning

10.1101/2020.03.19.998898 ◽

2020 ◽

Author(s):

Zhaoping Xiong ◽

Ziqiang Cheng ◽

Chi Xu ◽

Xinyuan Lin ◽

Xiaohong Liu ◽

...

Keyword(s):

Artificial Intelligence ◽

Drug Discovery ◽

Aqueous Solubility ◽

Training Data ◽

Small Data ◽

Sensitive Information ◽

Distributed Data ◽

High Quality ◽

Different Sources ◽

Striking Contrast

AbstractArtificial intelligence (AI) models usually require large amounts of high-quality training data, which is in striking contrast to the situation of small and biased data faced by current drug discovery pipelines. The concept of federated learning has been proposed to utilize distributed data from different sources without leaking sensitive information of these data. This emerging decentralized machine learning paradigm is expected to dramatically improve the success of AI-powered drug discovery. We here simulate the federated learning process with 7 aqueous solubility datasets from different sources, among which there are overlapping molecules with high or low biases in the recorded values. Beyond the benefit of gaining more data, we also demonstrate federated training has a regularization effect making it superior than centralized training on the pooled datasets with high biases. Further, two more cases are studied to test the usability of federated learning in drug discovery. Our work demonstrates the application of federated learning in predicting drug related properties, but also highlights its promising role in addressing the small data and biased data dilemma in drug discovery.

Download Full-text

Towards a supervised rescoring system for unstructured data bases used to build specialized dictionaries

Revista Facultad de Ingeniería ◽

10.19053/01211129.3161 ◽

2014 ◽

Vol 24 (38) ◽

pp. 97

Author(s):

Antonio Rico-Sulayes

Keyword(s):

Supervised Learning ◽

Learning Model ◽

Training Data ◽

Unstructured Data ◽

High Quality ◽

Data Bases ◽

Common Resource ◽

The Face ◽

Lexical Items ◽

Selection Of

<p align="justify">This article proposes the architecture for a system that uses previously learned weights to sort query results from unstructured data bases when building specialized dictionaries. A common resource in the construction of dictionaries, unstructured data bases have been especially useful in providing information about lexical items frequencies and examples in use. However, when building specialized dictionaries, whose selection of lexical items does not rely on frequency, the use of these data bases gets restricted to a simple provider of examples. Even in this task, the information unstructured data bases provide may not be very useful when looking for specialized uses of lexical items with various meanings and very long lists of results. In the face of this problem, long lists of hits can be rescored based on a supervised learning model that relies on previously helpful results. The allocation of a vast set of high quality training data for this rescoring system is reported here. Finally, the architecture of sucha system,an unprecedented tool in specialized lexicography, is proposed.</p>

Download Full-text

Learn to Train: Improving Training Data for a Neural Network to Detect Pecking Injuries in Turkeys

Animals ◽

10.3390/ani11092655 ◽

2021 ◽

Vol 11 (9) ◽

pp. 2655

Author(s):

Nina Volkmann ◽

Johannes Brünger ◽

Jenny Stracke ◽

Claudius Zelenka ◽

Reinhard Koch ◽

...

Keyword(s):

Neural Network ◽

Artificial Intelligence ◽

Automated Detection ◽

Training Data ◽

High Quality

This study aimed to develop a camera-based system using artificial intelligence for automated detection of pecking injuries in turkeys. Videos were recorded and split into individual images for further processing. Using specifically developed software, the injuries visible on these images were marked by humans, and a neural network was trained with these annotations. Due to unacceptable agreement between the annotations of humans and the network, several work steps were initiated to improve the training data. First, a costly work step was used to create high-quality annotations (HQA) for which multiple observers evaluated already annotated injuries. Therefore, each labeled detection had to be validated by three observers before it was saved as “finished”, and for each image, all detections had to be verified three times. Then, a network was trained with these HQA to assist observers in annotating more data. Finally, the benefit of the work step generating HQA was tested, and it was shown that the value of the agreement between the annotations of humans and the network could be doubled. Although the system is not yet capable of ensuring adequate detection of pecking injuries, the study demonstrated the importance of such validation steps in order to obtain good training data.

Download Full-text