Protein–DNA binding: complexities and multi-protein codes

Siggers, Trevor; Gordân, Raluca

doi:10.1093/nar/gkt1112

Cited by 218 publications

(208 citation statements)

References 130 publications

(195 reference statements)

Supporting

Mentioning

202

Contrasting

Order By: Relevance

“…The sequence specificities of a protein are most commonly characterized using position weight matrices 1 (PWMs), which are easy to interpret and can be scanned over a genomic sequence to detect potential binding sites. However, growing evidence indicates that sequence specificities can be more accurately captured by more complex techniques [2][3][4][5] . Recently, 'deep learning' has achieved record-breaking performance in a variety of information technology applications 6,7 .…”

Section: A N a Ly S I Smentioning

confidence: 99%

Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning

et al. 2015

View full text Add to dashboard Cite

Knowing the sequence specificities of DNA-and RNA-binding proteins is essential for developing models of the regulatory processes in biological systems and for identifying causal disease variants. Here we show that sequence specificities can be ascertained from experimental data with 'deep learning' techniques, which offer a scalable, flexible and unified computational approach for pattern discovery. Using a diverse array of experimental data and evaluation metrics, we find that deep learning outperforms other state-of-the-art methods, even when training on in vitro data and testing on in vivo data. We call this approach DeepBind and have built a stand-alone software tool that is fully automatic and handles millions of sequences per experiment. Specificities determined by DeepBind are readily visualized as a weighted ensemble of position weight matrices or as a 'mutation map' that indicates how variations affect binding within a specific sequence.DNA-and RNA-binding proteins play a central role in gene regulation, including transcription and alternative splicing. The sequence specificities of a protein are most commonly characterized using position weight matrices 1 (PWMs), which are easy to interpret and can be scanned over a genomic sequence to detect potential binding sites. However, growing evidence indicates that sequence specificities can be more accurately captured by more complex techniques 2-5 . Recently, 'deep learning' has achieved record-breaking performance in a variety of information technology applications 6,7 . We adapted deep learning methods to the task of predicting sequence specificities and found that they compete favorably with the state of the art. Our approach, called DeepBind, is based on deep convolutional neural networks and can discover new patterns even when the locations of patterns within sequences are unknown-a task for which traditional neural networks require an exorbitant amount of training data.There are several challenging aspects in learning models of sequence specificity using modern high-throughput technologies. First, the data come in qualitatively different forms. Protein binding microarrays (PBMs) 8 and RNAcompete assays 9 provide a specificity coefficient for each probe sequence, whereas chromatin immunoprecipitation (ChIP)-seq 10 provides a ranked list of putatively bound sequences of varying length, and HT-SELEX 11 generates a set of very high affinity sequences. Second, the quantity of data is large. A typical high-throughput experiment measures between 10,000 and 100,000 sequences, and it is computationally demanding to incorporate them all. Third, each data acquisition technology has its own artifacts, biases and limitations, and we must discover the pertinent specificities despite these unwanted effects. For example, ChIP-seq reads often localize to "hyper-ChIPable" regions of the genome near highly expressed genes 12 .DeepBind (Fig. 1) addresses the above challenges. (i) It can be applied to both microarray and sequencing data; (ii) it can learn from millions of...

show abstract

Section: A N a Ly S I Smentioning

confidence: 99%

Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning

et al. 2015

View full text Add to dashboard Cite

show abstract

“…[27][28][29][30] Interestingly, the substrate preferences of A3 proteins can be exchanged through the transfer of certain sequence regions. 31 In conjunction with other studies, this work established the role of loops 1, 3 and 7 in the process of DNA substrate recognition.…”

Section: Introductionmentioning

confidence: 99%

Determinants of Substrate Specificity in APOBEC3B

Wagner

Demir²,

Carpenter³

et al. 2018

Preprint

View full text Add to dashboard Cite

APOBEC3B (A3B) is a newly discovered driver of mutation in many cancers. We use computational tools to revert a recent crystal structure of an A3B construct to its native sequence, and run molecular dynamics simulations to study its underlying dynamics and substrate recognition mechanisms. The A3B oligonucleotide substrate simulations show a series of dynamic substrate protein contacts that correlate with previous work on A3B substrate selectivity. A second series of simulations in which the target cytosine nucleotide was computationally mutated from a deoxyribose to a ribose showed a change in sugar ring pucker, leading to a rearrangement of the binding site and revealing a potential intermediate in the binding pathway. Finally, apo simulations of A3B beginning in the open state experience a rapid and consistent closure of the binding site, reaching a conformation incompatible with substrate binding. These simulations agree with previous experimental studies, and we report the atomistic details of these events to further therapeutic studies on A3B. IntroductionThe APOBEC3 (A3) family of cytidine deaminases is a recently discovered endogenous source of mutation in cancer. 1,2 Previously, A3 proteins were studied in the context of their interactions with viruses and efforts were undertaken to discover small molecules that modulate the mutational activities of the virally restrictive APOBEC3G enzyme. 3,4 Recent studies have linked cancer progression and recurrence to A3 activity. 5-9 A3 proteins have specific substrate sequence preferences, and analyses of some cancer genomes have shown an enrichment of A3 mutation signatures. [10][11][12][13][14][15] Recent evidence suggests that APOBEC3B (A3B) is an important driver of tumor progression in the A3 family. 16,17

show abstract

“…The latest version of this technique, ChIP-exo, uses exonucleases to trim immunoprecipitated DNA fragments to precise locations of bound proteins, achieving single-base-pair precision (Rhee and Pugh 2011). Many factors influence TF binding locations in living cells, including chromatin structure, which strongly modulates DNA accessibility (Felsenfeld and Groudine 2003), and cooperative or competing interactions with other DNAbinding proteins (Siggers and Gordan 2014). However, the primary determinant of TF binding locations is intrinsic DNA sequence specificity: TFs can recognize and bind their cognate sites on the background of numerous competing sites available in the genome.…”

mentioning

confidence: 99%

A Biophysical Approach to Predicting Protein–DNA Binding Energetics

Locke

Morozov

2015

Genetics

View full text Add to dashboard Cite

Sequence-specific interactions between proteins and DNA play a central role in DNA replication, repair, recombination, and control of gene expression. These interactions can be studied in vitro using microfluidics, protein-binding microarrays (PBMs), and other high-throughput techniques. Here we develop a biophysical approach to predicting protein-DNA binding specificities from high-throughput in vitro data. Our algorithm, called BindSter, can model alternative DNA-binding modes and multiple protein species competing for access to DNA, while rigorously taking into account all sterically allowed configurations of DNA-bound factors. BindSter can be used with a hierarchy of protein-DNA interaction models of increasing complexity, including contributions of mononucleotides, dinucleotides, and longer words to the total protein-DNA binding energy. We observe that the quality of BindSter predictions does not change significantly as some of the energy parameters vary over a sizable range. To take this degeneracy into account, we have developed a graphical representation of parameter uncertainties called IntervalLogo. We find that our simplest model, in which each nucleotide in the binding site is treated independently, performs better than previous biophysical approaches. The extensions of this model, in which contributions of longer words are also considered, result in further improvements, underscoring the importance of higher-order effects in protein-DNA energetics. In contrast, we find little evidence of multiple binding modes for the transcription factors (TFs) and experimental conditions in our data set. Furthermore, there is limited consistency in predictions for the same TF based on microfluidics and PBM data.KEYWORDS gene regulation; protein-DNA interactions; thermodynamic modeling; transcription factor S EQUENCE-SPECIFIC interactions between proteins and genomic DNA control numerous cellular processes. An important class of DNA-binding proteins is transcription factors (TFs), which regulate expression of their target genes. Knowledge of in vivo binding locations of TFs is important for reconstructing regulatory networks of gene transcription and for understanding which factors control which genes. These locations can be mapped genome-wide, using chromatin immunoprecipitation followed by microarray hybridization (ChIP-chip) or sequencing (ChIP-seq) (Ren et al. 2000;Park 2009). The latest version of this technique, ChIP-exo, uses exonucleases to trim immunoprecipitated DNA fragments to precise locations of bound proteins, achieving single-base-pair precision (Rhee and Pugh 2011). Many factors influence TF binding locations in living cells, including chromatin structure, which strongly modulates DNA accessibility (Felsenfeld and Groudine 2003), and cooperative or competing interactions with other DNAbinding proteins (Siggers and Gordan 2014). However, the primary determinant of TF binding locations is intrinsic DNA sequence specificity: TFs can recognize and bind their cognate sites on the background of ...

show abstract

Protein–DNA binding: complexities and multi-protein codes

Cited by 218 publications

References 130 publications

Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning

Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning

Determinants of Substrate Specificity in APOBEC3B

A Biophysical Approach to Predicting Protein–DNA Binding Energetics

Contact Info

Product

Resources

About