Biology subjects

Novakovsky, G.

Publications and source records attributed to Novakovsky, G..

6 recordsLinked to original sources

bioRxiv2022

ExplaiNN: interpretable and transparent neural networks for genomics

Sequence-based deep learning models, particularly convolutional neural networks (CNNs), have shown superior performance on a wide range of genomic tasks. A key limitation of these models is the lack of interpretability, slowing down their adoption by the genomics community. Current approaches to model interpretation do not readily reveal how a model makes predictions, can be computationally intensive, and depend on the implemented architecture. Here, we introduce ExplaiNN, an adaptation of neural additive models[1] for genomic tasks wherein predictions are computed as a linear combination of multiple independent CNNs, each consisting of a single convolutional filter and fully connected layers. This approach brings together the expressiveness of CNNs with the interpretability of linear models, providing global (cell state level) as well as local (individual sequence level) biological insights into the data. We use ExplaiNN to predict transcription factor (TF) binding and chromatin accessibility states, demonstrating performance levels comparable to state-of-the-art methods, while providing a transparent view of the models predictions in a straightforward manner. Applied to de novo motif discovery, ExplaiNN identifies equivalent motifs to those obtained from specialized algorithms across a range of datasets. Finally, we present ExplaiNN as a plug-and-play platform in which pretrained TF binding models and annotated position weight matrices from reference databases can be easily combined. We expect that ExplaiNN will accelerate the adoption of deep learning by biological domain experts in their daily genomic sequence analyses.

genomics↗

bioRxiv2021

In silico discovery of small molecules for efficient stem cell differentiation into definitive endoderm

Improving methods for human embryonic stem cell differentiation represents a challenge in modern regenerative medicine research. Using drug repurposing approaches, we discover small molecules that regulate the formation of definitive endoderm. Among them are inhibitors of known processes involved in endoderm differentiation (mTOR, PI3K, and JNK pathways) and a new compound, with an unknown mechanism of action, capable of inducing endoderm formation in the absence of growth factors in the media. Optimization of the classical protocol by including this compound achieves the same differentiation efficiency with a 90% cost reduction. The gene expression profile induced by the compound suggests that it is an inhibitor of the MYC pathway. The proposed in silico procedure for candidate molecule selection has broad potential for improving stem cell differentiation protocols.

bioinformatics↗

bioRxiv2020

Biologically-relevant transfer learning improves transcription factor binding prediction

BackgroundDeep learning has proven to be a powerful technique for transcription factor (TF) binding prediction, but requires large training datasets. Transfer learning can reduce the amount of data required for deep learning, while improving overall model performance, compared to training a separate model for each new task. ResultsWe assess a transfer learning strategy for TF binding prediction consisting of a pre-training step, wherein we train a multi-task model with multiple TFs, and a fine-tuning step, wherein we initialize single-task models for individual TFs with the weights learned by the multi-task model, after which the single-task models are trained at a lower learning rate. We corroborate that transfer learning improves model performance, especially if in the pre-training step the multi-task model is trained with biologically-relevant TFs. We show the effectiveness of transfer learning for TFs with [~]500 ChIP-seq peak regions. Using model interpretation techniques, we demonstrate that the features learned in the pre-training step are refined in the fine-tuning step to resemble the binding motif of the target TF (i.e. the recipient of transfer learning in the fine-tuning step). Moreover, pre-training with biologically-relevant TFs allows single-task models in the fine-tuning step to learn features other than the motif of the target TF. ConclusionsOur results confirm that transfer learning is a powerful technique for TF binding prediction.

bioinformatics↗

bioRxiv2020

Petabase-scale sequence alignment catalyses viral discovery

Public databases contain a planetary collection of nucleic acid sequences, but their systematic exploration has been inhibited by a lack of efficient methods for searching this corpus, now exceeding multiple petabases and growing exponentially [1, 2]. We developed a cloud computing infrastructure, Serratus, to enable ultra-high throughput sequence alignment at the petabase scale. We searched 5.7 million biologically diverse samples (10.2 petabases) for the hallmark gene RNA dependent RNA polymerase, identifying well over 105 novel RNA viruses and thereby expanding the number of known species by roughly an order of magnitude. We characterised novel viruses related to coronaviruses and to hepatitis{delta} virus, respectively and explored their environmental reservoirs. To catalyse a new era of viral discovery, we established a free and comprehensive database of these data and tools. Expanding the known sequence diversity of viruses can reveal the evolutionary origins of emerging pathogens and improve pathogen surveillance for the anticipation and mitigation of future pandemics.

bioinformatics↗

bioRxiv2020

Repeated stimulation or tonic-signaling chimeric antigen receptors drive regulatory T cell exhaustion

Regulatory T cell (Treg) therapy is a promising approach to improve outcomes in transplantation and autoimmunity. In conventional T cell therapy, chronic stimulation can result in poor in vivo function, a phenomenon termed exhaustion. Whether or not Tregs are also susceptible to exhaustion, and if so, if this would limit their therapeutic effect, was unknown. We studied how two methods which induce conventional T cell exhaustion – repetitive stimulation or expression of a tonic-signaling chimeric antigen receptor (CAR) – affect human Tregs. With each repetitive polyclonal stimulation Tregs progressively acquired an exhausted phenotype, and became less suppressive in vitro. Tregs expressing a tonic-signaling CAR rapidly acquired an exhausted phenotype and had major changes in their transcriptome and metabolism. Although tonic-signaling CAR-Tregs remained stable and suppressive in vitro, they lost in vivo function, as tested in a model of xenogeneic graft-versus-host disease. The finding that human Tregs are susceptible to exhaustion has important implications for the design of Treg adoptive immunotherapy strategies.Competing Interest StatementMKL received research funding from TxCell, Bristol-Myers Squibb, Pfizer, Takeda and CRISPR Therapeutics for work unrelated to this report. CLM holds several patent applications in the area of CAR based immunotherapy, and is a founder of, holds equity in, and receives consulting fees from Lyell Immunopharma which develops CAR based therapies. EWW holds several patent applications in the area of CAR based immunotherapy and receives consulting fees from Lyell Immunopharma. CLM is a member of the Parker Institute for Cancer Immunotherapy, which supports the Stanford University Cancer Immunotherapy Program.AbbreviationsCARChimeric antigen receptorECARextracellular acidification rateOCARoxygen consumption rateTconvconventional T cellTregRegulatory T cellTStonic signalingTSDRRegulatory T-cell-Specific Demethylated RegionView Full Text

immunology↗

bioRxiv2019

Functional effects of chimeric antigen receptor co-receptor signaling domains in human Tregs

Antigen-specific regulatory T cells (Tregs) engineered with chimeric antigen receptor (CARs) are a potent immunosuppressive cellular therapy in multiple disease models. To date the majority of CAR Treg studies employed second generation CARs, encoding a CD28 or 4-1BB co-receptor signaling domain and CD3{zeta}, but it was not known if this CAR design was optimal for Tregs. Using an HLA-A2-specific CAR platform and human Tregs, we compared ten CARs with different co-receptor signaling domains and systematically tested their function. Tregs expressing a CAR encoding wild-type CD28 were markedly superior to all other CARs tested in an in vivo model of graft-versus-host disease. In vitro assays revealed stable expression of Helios and ability to suppress CD80 expression on DCs as key in vitro predictors of in vivo function. This comprehensive study of CAR signaling-domain variants in Tregs can be leveraged to optimize CAR design for use in antigen-specific Treg therapy.

immunology↗