Predicted Transcription Factor Binding Research Articles

MulTFBS: A Spatial-Temporal Network with Multichannels for Predicting Transcription Factor Binding Sites.

Revealing the mechanisms that influence transcription factor binding specificity is the key to understanding gene regulation. In previous studies, DNA double helix structure and one-hot embedding have been used successfully to design computational methods for predicting transcription factor binding sites (TFBSs). However, DNA sequence as a kind of biological language, the method of word embedding representation in natural language processing, has not been considered properly in TFBS prediction models. In our work, we integrate different types of features of DNA sequence to design a multichanneled deep learning framework, namely MulTFBS, in which independent one-hot encoding, word embedding encoding, which can incorporate contextual information and extract the global features of the sequences, and double helix three-dimensional structural features have been trained in different channels. To extract sequence high-level information effectively, in our deep learning framework, we select the spatial-temporal network by combining convolutional neural networks and bidirectional long short-term memory networks with attention mechanism. Compared with six state-of-the-art methods on 66 universal protein-binding microarray data sets of different transcription factors, MulTFBS performs best on all data sets in the regression tasks, with the average R2 of 0.698 and the average PCC of 0.833, which are 5.4% and 3.2% higher, respectively, than the suboptimal method CRPTS. In addition, we evaluate the classification performance of MulTFBS for distinguishing bound or unbound regions on TF ChIP-seq data. The results show that our framework also performs well in the TFBS classification tasks.

Predicting Transcription Factor Binding Sites with Deep Learning.

Prediction of binding sites for transcription factors is important to understand how the latter regulate gene expression and how this regulation can be modulated for therapeutic purposes. A consistent number of references address this issue with different approaches, Machine Learning being one of the most successful. Nevertheless, we note that many such approaches fail to propose a robust and meaningful method to embed the genetic data under analysis. We try to overcome this problem by proposing a bidirectional transformer-based encoder, empowered by bidirectional long-short term memory layers and with a capsule layer responsible for the final prediction. To evaluate the efficiency of the proposed approach, we use benchmark ChIP-seq datasets of five cell lines available in the ENCODE repository (A549, GM12878, Hep-G2, H1-hESC, and Hela). The results show that the proposed method can predict TFBS within the five different cell lines very well; moreover, cross-cell predictions provide satisfactory results as well. Experiments conducted across cell lines are reinforced by the analysis of five additional lines used only to test the model trained using the others. The results confirm that prediction across cell lines remains very high, allowing an extensive cross-transcription factor analysis to be performed from which several indications of interest for molecular biology may be drawn.

Benchmarking tools for transcription factor prioritization

Spatiotemporal regulation of gene expression is controlled by transcription factor (TF) binding to regulatory elements, resulting in a plethora of cell types and cell states from the same genetic information. Due to the importance of regulatory elements, various sequencing methods have been developed to localise them in genomes, for example using ChIP-seq profiling of the histone mark H3K27ac that marks active regulatory regions. Moreover, multiple tools have been developed to predict TF binding to these regulatory elements based on DNA sequence. As altered gene expression is a hallmark of disease phenotypes, identifying TFs driving such gene expression programs is critical for the identification of novel drug targets.In this study, we curated 84 chromatin profiling experiments (H3K27ac ChIP-seq) where TFs were perturbed through e.g., genetic knockout or overexpression. We ran nine published tools to prioritize TFs using these real-world data sets and evaluated the performance of the methods in identifying the perturbed TFs. This allowed the nomination of three frontrunner tools, namely RcisTarget, MEIRLOP and monaLisa. Our analyses revealed opportunities and commonalities of tools that will help to guide further improvements and developments in the field.

Deep learning model for predicting genetic diseases using DNA sequence data

The emergence of machine learning in the recent decade has excelled in determining new potential features and nonlinear relationships existing between the data derived from the DNA sequences of genetic diseases. Machine learning also enhances the process of handling data with maximum predicted variables compared to observations during the data mining process of prediction. In this context, our study presents a deep learning model for predicting Transcription Factor Binding Sites (TFBS) in DNA sequences, with a focus on features within genetic data associated with diseases. Transcription Factors (TFs) play a crucial role in modulating gene expression by binding to TFBS. The accurate prediction of TFBS is essential for understanding genome function and evolution. Thus, we develop an efficient deep learning model that considers TFBS prediction as a nucleotide-level binary classification task. In our proposed model, first we create an input matrix using the original DNA sequences. Next, we encode these DNA sequences using one-hot encoding, representing them as a sequence of numerical values. We then employ three convolutional layers, allowing our model to capture intricate patterns and motif features over a larger spatial range. To capture important features within the DNA sequence and to focus on them, we incorporate an attention layer. Finally, a dense layer, consisting of two fully connected layers and a dropout layer, calculates the probability of TF binding site occurrence based on the features learned by the proposed model. Our experimental results, using in-vivo datasets obtained from Chip-seq, demonstrate the superior performance of our proposed deep learning model in TFBS prediction compared to other existing state-of-the-art methods. The improvement in accuracy is due to additional layers of CNN and then an attention layer in the model. Thus, this result in a better performance of our approach in predicting the transcription factor binding sites and enhancing our understanding of gene regulation and genome function.

Omics-based construction of regulatory variants can be applied to help decipher pig liver-related traits

Genetic variants can influence complex traits by altering gene expression through changes to regulatory elements. However, the genetic variants that affect the activity of regulatory elements in pigs are largely unknown, and the extent to which these variants influence gene expression and contribute to the understanding of complex phenotypes remains unclear. Here, we annotate 90,991 high-quality regulatory elements using acetylation of histone H3 on lysine 27 (H3K27ac) ChIP-seq of 292 pig livers. Combined with genome resequencing and RNA-seq data, we identify 28,425 H3K27ac quantitative trait loci (acQTLs) and 12,250 expression quantitative trait loci (eQTLs). Through the allelic imbalance analysis, we validate two causative acQTL variants in independent datasets. We observe substantial sharing of genetic controls between gene expression and H3K27ac, particularly within promoters. We infer that 46% of H3K27ac exhibit a concomitant rather than causative relationship with gene expression. By integrating GWAS, eQTLs, acQTLs, and transcription factor binding prediction, we further demonstrate their application, through metabolites dulcitol, phosphatidylcholine (PC) (16:0/16:0) and published phenotypes, in identifying likely causal variants and genes, and discovering sub-threshold GWAS loci. We provide insight into the relationship between regulatory elements and gene expression, and the genetic foundation for dissecting the molecular mechanism of phenotypes.

Using methylation data to improve transcription factor binding prediction

ABSTRACT Modelling the regulatory mechanisms that determine cell fate, response to external perturbation, and disease state depends on measuring many factors, a task made more difficult by the plasticity of the epigenome. Scanning the genome for the sequence patterns defined by Position Weight Matrices (PWM) can be used to estimate transcription factor (TF) binding locations. However, this approach does not incorporate information regarding the epigenetic context necessary for TF binding. CpG methylation is an epigenetic mark influenced by environmental factors that is commonly assayed in human cohort studies. We developed a framework to score inferred TF binding locations using methylation data. We intersected motif locations identified using PWMs with methylation information captured in both whole-genome bisulfite sequencing and Illumina EPIC array data for six cell lines, scored motif locations based on these data, and compared with experimental data characterizing TF binding (ChIP-seq). We found that for most TFs, binding prediction improves using methylation-based scoring compared to standard PWM-scores. We also illustrate that our approach can be generalized to infer TF binding when methylation information is only proximally available, i.e. measured for nearby CpGs that do not directly overlap with a motif location. Overall, our approach provides a framework for inferring context-specific TF binding using methylation data. Importantly, the availability of DNA methylation data in existing patient populations provides an opportunity to use our approach to understand the impact of methylation on gene regulatory processes in the context of human disease.

Discovery of a non-canonical GRHL1 binding site using deep convolutional and recurrent neural networks

BackgroundTranscription factors regulate gene expression by binding to transcription factor binding sites (TFBSs). Most models for predicting TFBSs are based on position weight matrices (PWMs), which require a specific motif to be present in the DNA sequence and do not consider interdependencies of nucleotides. Novel approaches such as Transcription Factor Flexible Models or recurrent neural networks consequently provide higher accuracies. However, it is unclear whether such approaches can uncover novel non-canonical, hitherto unexpected TFBSs relevant to human transcriptional regulation.ResultsIn this study, we trained a convolutional recurrent neural network with HT-SELEX data for GRHL1 binding and applied it to a set of GRHL1 binding sites obtained from ChIP-Seq experiments from human cells. We identified 46 non-canonical GRHL1 binding sites, which were not found by a conventional PWM approach. Unexpectedly, some of the newly predicted binding sequences lacked the CNNG core motif, so far considered obligatory for GRHL1 binding. Using isothermal titration calorimetry, we experimentally confirmed binding between the GRHL1-DNA binding domain and predicted GRHL1 binding sites, including a non-canonical GRHL1 binding site. Mutagenesis of individual nucleotides revealed a correlation between predicted binding strength and experimentally validated binding affinity across representative sequences. This correlation was neither observed with a PWM-based nor another deep learning approach.ConclusionsOur results show that convolutional recurrent neural networks may uncover unanticipated binding sites and facilitate quantitative transcription factor binding predictions.

BTG2 Super-Enhancer Mutations Disrupt TFAP4 Binding and Deregulate BTG2 Expression in Diffuse Large B-Cell Lymphoma

Diffuse Large B-cell Lymphoma (DLBCL), the most frequent lymphoid malignancy, remains incurable in ~40% of patients. Coding-genome sequencing efforts identified several genes/pathways altered in this disease, as well as genetic subgroups of prognostic and therapeutic significance. However, the large non-coding portion of the genome remained largely unexplored. We recently identified a pervasive hypermutation mechanism targeting active super-enhancers (SEs) in >90% of DLBCLs and leading to dysregulation of multiple genes, including well-known lymphoma oncogenes (Bal et al., Nature 2022). This pervasive phenomenon exhibits hallmarks of AID activity, and we provided evidence of its oncogenic relevance by demonstrating that mutational hotspots in the BCL6, BCL2 and CXCR4 SEs impair the binding of specific transcriptional repressors, preventing their negative regulation and creating oncogenic dependencies in DLBCL cells. Here we aimed to define the pathogenic role of non-coding mutations targeting the intragenic SE (iSE) of the BTG2 gene, the second most commonly mutated in DLBCL (39% of samples analyzed, including 36/93 primary cases and 7/29 cell lines). BTG2 encodes a member of the B-cell translocation gene (BTG)/ Transducer of ErbB2.1(TOB) family involved in transcriptional co-activation and modulation of mRNA abundance. BTG2 is also targeted by somatic missense mutations in 6-11% of DLBCL samples, suggesting a major role in the pathogenesis of this disease, which has not been explored. In order to identify functionally relevant BTG2 SE mutations, we first screened an extended panel of 286 DLBCL samples for the presence of mutational clusters targeting predicted transcription factor binding motifs. We identified a 20bp sequence stretch that was significantly hypermutated in these tumors, as compared to other lymphoma types or to non-lymphoid tumors (n=64/286, 22%; vs 0/17 Burkitt Lymphomas; 18/284 Follicular Lymphomas; 1/240 Chronic Lymphocytic Leukemias, and 0/1596 other tumors; p<0.0001, Fisher's exact test). In particular, mutations of the C at position +904 to G or T were exclusively detected in 11% DLBCL samples (32/286, p<0.0001, Fisher's exact test). To define the consequences of BTG2 C904 mutations, we then investigated the impact of reverting the mutation to the wild type nucleotide in 2 mutant DLBCL cell lines, using the CRISPR-Cas9 technology. Correction of the mutation led to a significant counter selection of the corrected clones (~85% reduction in a clone recovery assay, compared to control; p<0.001, Fisher's exact test) and to reduced BTG2 mRNA expression (p<0.01, one-way ANOVA), suggesting oncogenic addiction. In silico motif prediction and in vitro DNA-binding assays using mutant vs wild type oligonucleotides in two different DLBCL cell lines, followed by mass spectrometry, identified the Transcription Factor AP4 (TFAP4) as a protein specifically bound to the wild type, but not to the mutant sequences. TFAP4 is an important regulator of B-cell proliferation and cell fate decisions, which can function as a transcriptional activator or repressor in germinal center B-cells and acts downstream of/in parallel with c-MYC. We confirmed that the C904G and C904T SE mutation impaired TFAP4 binding in isogenic DLBCL cell lines by ChIP-qPCR and led to increased BTG2 expression (p<0.001, one-way ANOVA). Moreover, loss of TFAP4 in CRISPR-Cas9 edited DLBCL cells led to higher BTG2 expression levels, documenting TFAP4 as a direct negative regulator of BTG2 expression (p<0.001, one-way ANOVA). Consistently, single cell analysis of normal GC B cells revealed an inverse correlation between the expression pattern of TFAP4 and BTG2 in cells primed to undergo plasma cell differentiation vs cells committed to dark zone re-entry, suggesting that these two genes may function as a molecular switch in the cell fate decision of GC B cells. The programs modulated by BTG2 and TFAP4 in the GC, and disrupted in tumors carrying deregulated BTG2 alleles, will be presented, as defined by RNA-seq analysis of 3 isogenic GC-derived cell lines followed by interrogation of over 100 clinically annotated primary DLBCL biopsies. Together, these data identify TFAP4 as a novel regulator of BTG2 and reveal a previously unappreciated mechanism by which mutations in the BTG2 SE disrupt this circuit and contribute to lymphomagenesis through deregulating BTG2 expression.

Characterizing genetic variation in the regulation of the ER stress response through computational and cis-eQTL analyses.

Misfolded proteins in the endoplasmic reticulum (ER) elicit the ER stress response, a large transcriptional response driven by 3 well-characterized transcription factors (TFs). This transcriptional response is variable across different genetic backgrounds. One mechanism in which genetic variation can lead to transcriptional variability in the ER stress response is through altered binding and activity of the 3 main TFs: XBP1, ATF6, and ATF4. This work attempts to better understand this mechanism by first creating a computational pipeline to identify potential binding sites throughout the human genome. We utilized GTEx data sets to identify cis-eQTLs that fall within predicted TF binding sites (TFBSs). We also utilized the ClinVar database to compare the number of pathogenic vs benign variants at different positions of the binding motifs. Finally, we performed a cis-eQTL analysis on human cell lines experiencing ER stress to identify cis-eQTLs that regulate the variable ER stress response. The majority of these cis-eQTLs are unique to a given condition: control or ER stress. Some of these stress-specific cis-eQTLs fall within putative binding sites of the 3 main ER stress response TFs, providing a potential mechanism by which these cis-eQTLs might be impacting gene expression under ER stress conditions through altered TF binding. This study represents the first cis-eQTL analysis on human samples experiencing ER stress and is a vital step toward identifying the genetic components responsible for the variable ER stress response.

Integrative CUT&Tag-RNA-Seq analysis of histone variant macroH2A1-dependent orchestration of human induced pluripotent stem cellreprogramming.

Aim: Humaninduced pluripotent stem cells (iPSCs) are inefficiently derived from somatic cells by overexpression of defined transcription factors. Overexpression of H2A histone variant macroH2A1.1, but not macroH2A1.2, leads to increased iPSCreprogrammingby unclear mechanisms. Materials & methods: Cleavage under targets and tagmentation (CUT&Tag) allows robust epigenomic profiling of a low cell number. We performed an integrative CUT&Tag-RNA-Seq analysis of macroH2A1-dependent orchestration of iPSCs reprogramming using human endothelial cells. Results: We demonstrate wider genome occupancy, predicted transcription factors binding, and gene expression regulated by macroH2A1.1 during reprogramming, compared to macroH2A1.2. MacroH2A1.1, previously associated with neurodegenerative pathologies, specifically activated ectoderm/neural processes. Conclusion: CUT&Tag and RNA-Seq data integration is a powerful tool to investigate the epigenetic mechanisms occurring during cell reprogramming.

Using Fully Convolutional Network to Locate Transcription Factor Binding Sites Based on DNA Sequence and Conservation Information.

Transcription factors (TFs) play a part in gene expression. TFs can form complex gene expression regulation system by combining with DNA. Thereby, identifying the binding regions has become an indispensable step for understanding the regulatory mechanism of gene expression. Due to the great achievements of applying deep learning (DL) to computer vision and language processing in recent years, many scholars are inspired to use these methods to predict TF binding sites (TFBSs), achieving extraordinary results. However, these methods mainly focus on whether DNA sequences include TFBSs. In this paper, we propose a fully convolutional network (FCN) coupled with refinement residual block (RRB) and global average pooling layer (GAPL), namely FCNARRB. Our model could classify binding sequences at nucleotide level by outputting dense label for input data. Experimental results on human ChIP-seq datasets show that the RRB and GAPL structures are very useful for improving model performance. Adding GAPL improves the performance by 9.32% and 7.61% in terms of IoU (Intersection of Union) and PRAUC (Area Under Curve of Precision and Recall), and adding RRB improves the performance by 7.40% and 4.64%, respectively. In addition, we find that conservation information can help locate TFBSs.

Molecular determinants of the response of cancer cells towards geldanamycin and its derivatives

Geldanamycin is an ansamycin-derivative of a benzoquinone isolated from Streptomyces hygroscopicus. It inhibits tyrosine kinases and heat shock protein 90 (HSP90). Geldanamycin and 11 derivatives were subjected to molecular docking to HSP90, and 17-desmethoxy-17-N,N-dimethylamino-geldanamycin (17-DMAG) was the compound with the highest binding affinity (−7.73 ± 0.12 kcal/mol) and the lowest inhibition constant (2.16 ± 0.49 μM). Therefore, 17-DMAG was selected for further experiments in comparison to geldanamycin. Multidrug resistance (MDR) represents a major problem for successful cancer therapy. We tested geldanamycin and 17-DMAG against various drug-resistant cancer cell lines. Although geldanamycin and 17-DMAG inhibited the proliferation in all cell lines tested, multidrug-resistant P-glycoprotein-overexpressing CEM/ADR5000 cells were cross-resistant, ΔEGFR-overexpressing tumor cells and p53 knockout cells were sensitive to these two compounds. COMPARE and hierarchical cluster analyses were performed, and 60 genes were identified to predict the sensitivity or resistance of 59 NCI tumor cell lines towards geldanamycin and 17-DMAG. The distribution of cell lines according to their mRNA expression profiles indicated sensitivity or resistance to both compounds with statistical significance. Moreover, bioinformatic tools were used to study possible mechanisms of action of geldanamycin and 17-DMAG. Galaxy Cistrome analyses were carried out to predict transcription factor binding motifs in the promoter regions of the candidate genes. Interestingly, the NF-ĸB DNA binding motif (Rel) was identified as the top transcription factor. Furthermore, these 60 genes were subjected to Ingenuity Pathway Analysis (IPA) to study the signaling pathway interactions of these genes. Interestingly, IPA also revealed the NF-ĸB pathway as the top network among these genes. Finally, NF-ĸB reporter assays confirmed the bioinformatic prediction, and both geldanamycin and 17-DMAG significantly inhibited NF-κB activity after exposure for 24 h. In conclusion, geldanamycin and 17-DMAG exhibited cytotoxic activity against different tumor cell lines. Their activity was not restricted to HSP90 but indicated an involvement of the NF-KB pathway.

Analysis of Functional Promoter of Camel FGF21 Gene and Identification of Small Compounds Targeting FGF21 Protein.

The fibroblast growth factor 21 (FGF21) gene plays an important role in the mechanism of glucose and lipid metabolism and is a promising therapeutic target for metabolic disease. Camels display a unique regulation characteristic of glucose and lipid metabolism, endowing them with the ability to adapt to survive drought and chronic hunger. However, the knowledge about the camel FGF21 gene regulation and its differences between humans and mice is still limited. In this study, camel FGF21 gene promoter was obtained for ~2000 bp upstream of the transcriptional start site (TSS). Bioinformatics analysis showed that the proximal promoter region sequences near the TSS between humans and camels have high similarity. Two potential core active regions are located in the -445-612 bp region. In addition, camel FGF21 promoter contains three CpG islands (CGIs), located in the -435~-1168 bp regions, significantly more and longer than in humans and mice. The transcription factor binding prediction showed that most transcription factors, including major functional transcription factors, are the same in different species although the binding site positions in the promoter are different. These results indicated that the signaling pathways involved in FGF21 gene transcription regulation are conservative in mammals. Truncated fragments recombinant vectors and luciferase reporter assay determined that camel FGF21 core promoter is located within the 800 bp region upstream of the TSS and an enhancer may exist between the -1000 and -2000 bp region. Combining molecular docking and in silico ADMET druggability prediction, two compounds were screened as the most promising candidate drugs specifically targeting FGF21. This study expanded the functions of these small molecules and provided a foundation for drug development targeting FGF21.

HAMPLE: deciphering TF-DNA binding mechanism in different cellular environments by characterizing higher-order nucleotide dependency.

Transcription factor (TF) binds to conservative DNA binding sites in different cellular environments and development stages by physical interaction with interdependent nucleotides. However, systematic computational characterization of the relationship between higher-order nucleotide dependency and TF-DNA binding mechanism in diverse cell types remains challenging. Here, we propose a novel multi-task learning framework HAMPLE to simultaneously predict TF binding sites (TFBS) in distinct cell types by characterizing higher-order nucleotide dependencies. Specifically, HAMPLE first represents a DNA sequence through three higher-order nucleotide dependencies, including k-mer encoding, DNA shape and histone modification. Then, HAMPLE employs the customized gate control and the channel attention convolutional architecture to further capture cell-type-specific and cell-type-shared DNA binding motifs and epigenomic languages. Finally, HAMPLE exploits the joint loss function to optimize the TFBS prediction for different cell types in an end-to-end manner. Extensive experimental results on seven datasets demonstrate that HAMPLE significantly outperforms the state-of-the-art approaches in terms of auROC. In addition, feature importance analysis illustrates that k-mer encoding, DNA shape and histone modification have predictive power for TF-DNA binding in different cellular environments and are complementary to each other. Furthermore, ablation study and interpretable analysis validate the effectiveness of the customized gate control and the channel attention convolutional architecture in characterizing higher-order nucleotide dependencies. The source code is available at https://github.com/ZhangLab312/Hample. Supplementary data are available at Bioinformatics online.

Differential Motif and Transcription Factor Binding Predictions Link to ULK4 Gene Polymorphisms and Associated with Essential Hypertension in Caucasian Females

Differential Motif and Transcription Factor Binding Predictions Link to ULK4 Gene Polymorphisms and Associated with Essential Hypertension in Caucasian Females

Xcore: an R package for inference of gene expression regulators

BackgroundElucidating the Transcription Factors (TFs) that drive the gene expression changes in a given experiment is a common question asked by researchers. The existing methods rely on the predicted Transcription Factor Binding Site (TFBS) to model the changes in the motif activity. Such methods only work for TFs that have a motif and assume the TF binding profile is the same in all cell types.ResultsGiven the wealth of the ChIP-seq data available for a wide range of the TFs in various cell types, we propose that gene expression modeling can be done using ChIP-seq “signatures” directly, effectively skipping the motif finding and TFBS prediction steps. We present xcore, an R package that allows TF activity modeling based on ChIP-seq signatures and the user's gene expression data. We also provide xcoredata a companion data package that provides a collection of preprocessed ChIP-seq signatures. We demonstrate that xcore leads to biologically relevant predictions using transforming growth factor beta induced epithelial-mesenchymal transition time-courses, rinderpest infection time-courses, and embryonic stem cells differentiated to cardiomyocytes time-course profiled with Cap Analysis Gene Expression.Conclusionsxcore provides a simple analytical framework for gene expression modeling using linear models that can be easily incorporated into differential expression analysis pipelines. Taking advantage of public ChIP-seq databases, xcore can identify meaningful molecular signatures and relevant ChIP-seq experiments.

Parkinson’s disease-associated 18 bp promoter variant of DJ-1 alters REST binding and regulates its expression

Parkinson’s disease (PD) is a progressive neurodegenerative disorder with a complex etiology. Presence of autosomal mutations in PARK7/DJ-1 gene has been associated with early-onset PD. Growing evidence has suggested that DJ-1 acts as a putative sensor of oxidative stress. Reduced levels of DJ-1 protein have been reported in the cerebrospinal fluid of sporadic PD patients. Several case-control association studies have identified DJ-1 g.168_185del (rs200968609) variants conferring susceptibility towards PD pathogenesis. Similarly, among the PD patients in eastern India, the deletion allele (g.168_185) of this DJ-1 promoter polymorphism was found to be associated with PD. Hence, we aimed to find out the functional contribution of this promoter variant of DJ-1 in PD pathogenesis. The expression of DJ-1 was observed to be significantly reduced in the presence of both deletion and duplication sequences as identified from the luciferase promoter activity assay. The transcription factor binding prediction tool identified DJ-1 promoter 18 bp insertion polymorphism as the only binding partner of REST (RE1 Silencing Transcription Factor). Transient Chromatin Immuno-precipitation (ChIP) assay further confirmed this prediction. Previous reports have highlighted the role of REST in regulating the expression of stress-responsive genes. Our study has identified the functional involvement of DJ-1 promoter variants and REST-mediated regulation of DJ-1 expression in PD pathogenesis.

NetTIME: a multitask and base-pair resolution framework for improved transcription factor binding site prediction.

MotivationMachine learning models for predicting cell-type-specific transcription factor (TF) binding sites have become increasingly more accurate thanks to the increased availability of next-generation sequencing data and more standardized model evaluation criteria. However, knowledge transfer from data-rich to data-limited TFs and cell types remains crucial for improving TF binding prediction models because available binding labels are highly skewed towards a small collection of TFs and cell types. Transfer prediction of TF binding sites can potentially benefit from a multitask learning approach; however, existing methods typically use shallow single-task models to generate low-resolution predictions. Here, we propose NetTIME, a multitask learning framework for predicting cell-type-specific TF binding sites with base-pair resolution.ResultsWe show that the multitask learning strategy for TF binding prediction is more efficient than the single-task approach due to the increased data availability. NetTIME trains high-dimensional embedding vectors to distinguish TF and cell-type identities. We show that this approach is critical for the success of the multitask learning strategy and allows our model to make accurate transfer predictions within and beyond the training panels of TFs and cell types. We additionally train a linear-chain conditional random field (CRF) to classify binding predictions and show that this CRF eliminates the need for setting a probability threshold and reduces classification noise. We compare our method’s predictive performance with two state-of-the-art methods, Catchitt and Leopard, and show that our method outperforms previous methods under both supervised and transfer learning settings.Availability and implementationNetTIME is freely available at https://github.com/ryi06/NetTIME and the code is also archived at https://doi.org/10.5281/zenodo.6994897.Supplementary information Supplementary data are available at Bioinformatics online.

P604: CATALASE EXPRESSION IN LEUKEMIA CELLS IS CONTROLLED BY GENETIC AND EPIGENETIC MECHANISMS

Background: Chronic lymphocytic leukemia (CLL) is the most prevalent form of leukemia in Western countries. It is an incurable disease characterized by an extremely variable clinical course and response to treatment. We have recently shown that high catalase (CAT) expression identifies a more aggressive clinical behavior in CLL. However, molecular mechanisms controlling catalase expression in leukemia cells are still poorly characterized. The rs1001179 single nucleotide polymorphism (SNP) as well as DNA methylation in the CAT promoter have been shown to regulate CAT expression in various cell types. Aims: With the aim to characterize regulatory mechanisms underlying differential expression of catalase in CLL, in this study we investigated the role of rs1001179 SNP and CpG Island II methylation encompassing this SNP in the regulation of CAT expression. Methods: Peripheral blood mononuclear cells (PBMC) from 55 CLL patients and 50 healthy donors (HD) were used for the study. CAT mRNA levels were measured using quantitative reverse transcription polymerase chain reaction (qRT-PCR). DNA samples were genotyped for the rs1001179 SNP using Restriction Fragment Length Polymorphism (RFLP)-PCR. Bioinformatic analysis was used to predict transcription factor (TF) binding to SNP sequences. Chromatin Immunoprecipitation (ChIP)-qPCR was performed to validate predicted TF binding sites. Methylation levels of CpG sites within the CAT promoter was determined by pyrosequencing of bisulfite-converted DNA. Results: The rs1001179 SNP genotyping showed that CLL cells harboring the T allele exhibited a significantly higher catalase expression compared with cells bearing the CC genotype. Interestingly, bioinformatic analysis predicted that the SNP provide distinct binding sites for various TF, including ETS-1 and GR-β. ChIP assay confirmed that CAT promoter harboring the T -but not C- allele was accessible to ETS-1 and GR-β, but not to the other analyzed TFs. In addition, CLL cells exhibited lower methylation levels within the CAT promoter compared with HD B cells, in line with the higher catalase mRNA and protein levels expressed by CLL versus HD B cells. Moreover, methylation levels negatively correlated with CAT expression in CLL cells. Remarkably, inhibition of methyltransferase activity in leukemic cells induced a significant increase of CAT mRNA and protein levels, thus showing that DNA methylation could control catalase expression in CLL. Moreover, expression of DNA methyl transferase 1 (DNMT1) resulted significantly reduced in CLL cells compared with HD B cells and inversely correlated with CAT expression in CLL, thus suggesting that differences in methylation levels underlying catalase expression could be driven by the DNMT1 enzyme. Finally, modeling analysis showed that the CT/TT genotypes exhibited a lower methylation and a higher CAT expression level, suggesting that the rs1001179 T allele and methylation cooperate in inducing CAT gene expression. Summary/Conclusion: Our data show that SNP as well as methylation of the catalase promoter are involved in controlling catalase expression in CLL. The key advance of this study is to provide new insights into the regulatory mechanisms underlying differential expression of catalase in leukemic cells, which is of clinical relevance in CLL. Moreover, our study may form the basis for future challenges aimed at developing combinatorial therapies targeting catalase regulatory pathways.

Virtual ChIP-seq: predicting transcription factor binding by learning from the transcriptome

Existing methods for computational prediction of transcription factor (TF) binding sites evaluate genomic regions with similarity to known TF sequence preferences. Most TF binding sites, however, do not resemble known TF sequence motifs, and many TFs are not sequence-specific. We developed Virtual ChIP-seq, which predicts binding of individual TFs in new cell types, integrating learned associations with gene expression and binding, TF binding sites from other cell types, and chromatin accessibility data in the new cell type. This approach outperforms methods that predict TF binding solely based on sequence preference, predicting binding for 36 TFs (MCC>0.3).