The DOI matches its Crossref record; 161 of the 163 references it cites were found in a registry or an index, and 1 are not checked yet.
Base-resolution models of transcription-factor binding reveal soft motif syntax
Open accessPublished version of a preprint
Abstract
The arrangement (syntax) of transcription factor (TF) binding motifs is an important part of the cis-regulatory code, yet remains elusive. We introduce a deep learning model, BPNet, that uses DNA sequence to predict base-resolution chromatin immunoprecipitation (ChIP)-nexus binding profiles of pluripotency TFs. We develop interpretation tools to learn predictive motif representations and identify soft syntax rules for cooperative TF binding interactions. Strikingly, Nanog preferentially binds with helical periodicity, and TFs often cooperate in a directional manner, which we validate using clustered regularly interspaced short palindromic repeat (CRISPR)-induced point mutations. Our model represents a powerful general approach to uncover the motifs and syntax of cis-regulatory sequences in genomics data.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“The BPNet architecture has been used widely in genomics and represents a very strong default architecture for modeling small-sized datasets from scratch through supervised learning.”
From Nucleotide Transformer: building and evaluating robust foundation models for human genomics (2025) · cites this paper as [9]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Machine-learning models trained to predict function from DNA sequences have been successful at characterizing regulatory syntax and interpreting genetic variant effects.”
From Predicting RNA-seq coverage from DNA sequence as a unifying model of gene regulation (2025) · cites this paper as [7]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Thus far, such models have focused on assays in which measured activity is proportional to local sequencing read counts.”
From Predicting RNA-seq coverage from DNA sequence as a unifying model of gene regulation (2025) · cites this paper as [7]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Finer resolution has potential benefits when the data support it15, but would extend the sequence length entering the quadratic complexity MHA and make the model engineering intractable on currently available hardware.”
From Effective gene expression prediction from sequence by integrating long-range interactions (2021) · cites this paper as [15]
The cited source says what the claim attributes to it.
In the citing paper
“Additionally, a deep learning algorithm applied to binding of pluripotency TFs at base resolution in mouse embryonic stem cells (ESCs) suggested a “soft motif syntax” whereby TFs help each other to bind directly to the enhancer with a lenient distance-dependency between binding sites [98].”
From Transcriptional enhancers and their communication with gene promoters (2021) · cites this paper as [98]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
VMD: visual molecular dynamics
Humphrey, W., Dalke, A., & Schulten, K. (1996) · Journal of Molecular Graphics · doi:10.1016/0263-7855(96)00018-5 · Cited by 66,952
The Sequence Alignment/Map format and SAMtools
Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G., Durbin, R., & 1000 Genome Project Data Processing Subgroup (2009) · Bioinformatics · doi:10.1093/bioinformatics/btp352 · Cited by 62,041
Fast and accurate short read alignment with Burrows-Wheeler transform
Li, H., & Durbin, R. (2009) · Bioinformatics · doi:10.1093/bioinformatics/btp324 · Cited by 58,607
Cutadapt removes adapter sequences from high-throughput sequencing reads
Martin, M. (2011) · EMBnet.journal · doi:10.14806/ej.17.1.200 · Cited by 33,889
Kimura, M. (1980) · Journal of Molecular Evolution · doi:10.1007/bf01731581 · Cited by 29,778
BEDTools: a flexible suite of utilities for comparing genomic features
Quinlan, A. R., & Hall, I. M. (2010) · Bioinformatics · doi:10.1093/bioinformatics/btq033 · Cited by 27,437
Ultrafast and memory-efficient alignment of short DNA sequences to the human genome.
Langmead, B., Trapnell, C., Pop, M., & Salzberg, S. L. (2009) · Genome Biology · doi:10.1186/gb-2009-10-3-r25 · Cited by 21,849
Hierarchical Grouping to Optimize an Objective Function
Ward, J. H. (1963) · Journal of the American Statistical Association · doi:10.1080/01621459.1963.10500845 · Cited by 21,341
Research works printed but not linked: 1
- Shrikumar, A. et al. TF-MoDISco v0.4.2.2-alpha: technical note. Preprint at arXiv https://arxiv.org/abs/1811.00416 (2018). · arXiv
Cited by
Sorted by citations
JASPAR 2022: the 9th release of the open-access database of transcription factor binding profiles
Castro-Mondragon, J. A., Riudavets-Puig, R., Rauluseviciute, I., Berhanu Lemma, R., Turchi, L., Blanc-Mathieu, R., Lucas, J., Boddie, P., Khan, A., Manosalva Pérez, N., et al. (2022) · Nucleic Acids Research · doi:10.1093/nar/gkab1113 · Cited by 1,935
“A novel generation of computational models based on machine learning approaches such as deep learning are arising (68,69).”
Effective gene expression prediction from sequence by integrating long-range interactions
Avsec, Ž., Agarwal, V., Visentin, D., Ledsam, J. R., Grabska-Barwinska, A., Taylor, K. R., Assael, Y., Jumper, J., Kohli, P., & Kelley, D. R. (2021) · Nature Methods · doi:10.1038/s41592-021-01252-x · Cited by 1,326
“Machine learning success depends on the training data, so increasing the resolution and quality of the target tracks 14 , and curating data from additional organisms 2 , would likely boost performance.”
BackgroundSelf-citationJASPAR 2024: 20th anniversary of the open-access database of transcription factor binding profiles
Rauluševičiūtė, I., Puig, R. R., Blanc‐Mathieu, R., Castro-Mondragon, J. A., Ferenc, K., Kumar, V., Lemma, R. B., Lucas, J., Chèneby, J., Baranas̆ić, D., et al. (2023) · Nucleic Acids Research · doi:10.1093/nar/gkad1059 · Cited by 1,110
“Mostly based on convolutional neural networks, deep learning models are now considered state-of-the-art to accurately model data generated from genomic assays such as ChIP-seq, ChIP-nexus, or A T AC-seq ( 47– 49 ).”
MethodNucleotide Transformer: building and evaluating robust foundation models for human genomics
Dalla-Torre, H., Gonzalez, L., Mendoza-Revilla, J., Lopez Carranza, N., Grzywaczewski, A. H., Oteri, F., Dallago, C., Trop, E., de Almeida, B. P., Sirelkhatim, H., et al. (2025) · Nature Methods · doi:10.1038/s41592-024-02523-z · Cited by 772
“In this context, specialized deep-learning (DL) models have been trained to uncover meaningful patterns of DNA 9–12 .”
InfluentialGENCODE: reference annotation for the human and mouse genomes in 2023
Frankish, A., Sala, S. C., Diekhans, M., Jungreis, I., Loveland, J. E., Mudge, J. M., Sisu, C., Wright, J. C., Arnan, C., Barnes, I., et al. (2022) · Nucleic Acids Research · doi:10.1093/nar/gkac1071 · Cited by 671
Self-citationWILDS: A Benchmark of in-the-Wild Distribution Shifts
Koh, P. W., Sagawa, S., Marklund, H., Xie, S. M., Zhang, M., Balsubramani, A., Hu, W., Yasunaga, M., Phillips, R. L., Gao, I., et al. (2020) · arXiv · doi:10.48550/arxiv.2012.07421 · Cited by 518
Self-citationAlphaGenome: advancing regulatory variant effect prediction with a unified DNA sequence model
Avsec, Ž., Latysheva, N., Cheng, J., Novati, G., Taylor, K. R., Ward, T., Bycroft, C., Nicolaisen, L., Arvaniti, E., Pan, J., et al. (2025) · doi:10.1101/2025.06.25.661532 · Cited by 395
Self-citationHow to build the virtual cell with artificial intelligence: Priorities and opportunities
Bunne, C., Roohani, Y., Rosen, Y., Gupta, A., Zhang, X., Roed, M., Alexandrov, T., AlQuraishi, M., Brennan, P., Burkhardt, D. B., et al. (2024) · Cell · doi:10.1016/j.cell.2024.11.015 · Cited by 393
“147 Despite their extensive utility, CNNs are increasingly being supplemented or replaced by vision transformer models, 72 which leverage self-attention mechanisms to process entire images in parallel.”
Obtaining genetics insights from deep learning via explainable artificial intelligence
Novakovsky, G., Dexter, N., Libbrecht, M. W., Wasserman, W. W., & Mostafavi, S. (2022) · Nature Reviews Genetics · doi:10.1038/s41576-022-00532-2 · Cited by 415
Advancing regulatory variant effect prediction with AlphaGenome
Avsec, Ž., Latysheva, N., Cheng, J., Novati, G., Taylor, K. R., Ward, T., Bycroft, C., Nicolaisen, L., Arvaniti, E., Pan, J., et al. (2026) · Nature · doi:10.1038/s41586-025-10014-0 · Cited by 373
Deciphering the multi-scale, quantitative cis-regulatory code
Kim, S., & Wysocka, J. (2023) · Molecular Cell · doi:10.1016/j.molcel.2022.12.032 · Cited by 302
A sequence-based global map of regulatory activity for deciphering human genetics
Chen, K. M., Wong, A. K., Troyanskaya, O. G., & Zhou, J. (2022) · Nature Genetics · doi:10.1038/s41588-022-01102-2 · Cited by 269
Transcription factors: Bridge between cell signaling and gene regulation
Weidemüller, P., Kholmatov, M. M., Petsalaki, E., & Zaugg, J. B. (2021) · PROTEOMICS · doi:10.1002/pmic.202000034 · Cited by 295
DeepSTARR predicts enhancer activity from DNA sequence and enables the de novo design of enhancers
de Almeida, B. P., Reiter, F., Pagani, M., & Stark, A. (2021) · doi:10.1101/2021.10.05.463203 · Cited by 284
Predicting RNA splicing from DNA sequence using Pangolin
Zeng, T., & Li, Y. I. (2022) · Genome Biology · doi:10.1186/s13059-022-02664-4 · Cited by 283
“Using sequences frommultiple species [18] and quantitative over binary labels [2, 19] have been shown to substantially improve prediction in some applications but have not yet been tested, to the best of our knowledge, in the context of RNA splicing.”
Methodde Almeida, B. P., Reiter, F., Pagani, M., & Stark, A. (2022) · Nature Genetics · doi:10.1038/s41588-022-01048-5 · Cited by 283
“49 ).”
BackgroundMethodInfluentialTranscription factors interact with RNA to regulate genes
Oksuz, O., Henninger, J. E., Warneford-Thomson, R., Zheng, M. M., Erb, H., Vancura, A., Overholt, K. J., Hawken, S. W., Banani, S. F., Lauman, R., et al. (2023) · Molecular Cell · doi:10.1016/j.molcel.2023.06.012 · Cited by 275
“TFs typically contain DNA-binding domains that recognize specific sequences and multiple TFs collectively bind to enhancers and promoter-proximal regions of genes (Avsec et al., 2021; Panne et al., 2007).”
BackgroundThe evolution, evolvability and engineering of gene regulatory DNA
Vaishnav, E. D., de Boer, C. G., Molinet, J., Yassour, M., Fan, L., Adiconis, X., Thompson, D. A., Levin, J. Z., Cubillos, F. A., & Regev, A. (2022) · Nature · doi:10.1038/s41586-022-04506-6 · Cited by 292
“A deep neural network model 20,21,23,61–69 with convolutional layers was constructed and used for designing sequences with high and low expression (Fig.”
MethodPredicting RNA-seq coverage from DNA sequence as a unifying model of gene regulation
Linder, J., Srivastava, D., Yuan, H., Agarwal, V., & Kelley, D. R. (2025) · Nature Genetics · doi:10.1038/s41588-024-02053-6 · Cited by 265
“We used a Poisson loss function, but decomposed the loss analogous to BPnet to separate magnitude and shape terms [7].”
Artificial intelligence in cancer research, diagnosis and therapy
Elemento, O., Leslie, C., Lundin, J., & Tourassi, G. (2021) · Nature Reviews Cancer · doi:10.1038/s41568-021-00399-1 · Cited by 264
“A beautiful recent study in this vein introduced BPNet 13 , a model for prediction of nucleotide-resolution transcription factor occupancy profiles and learning transcription factor ‘motif grammars’, embeddings — mappings of high-dimensional input data to a lower-dimensional ‘bottleneck’ or latent space — bringing new…”
BackgroundSequence determinants of human gene regulatory elements
Sahu, B., Hartonen, T., Pihlajamaa, P., Wei, B., Dave, K., Zhu, F., Kaasinen, E., Lidschreiber, K., Lidschreiber, M., Daub, C. O., et al. (2022) · Nature Genetics · doi:10.1038/s41588-021-01009-4 · Cited by 198
Adipose tissue at single-cell resolution
Maniyadath, B., Zhang, Q., Gupta, R. K., & Mandrup, S. (2023) · Cell Metabolism · doi:10.1016/j.cmet.2023.02.002 · Cited by 218
Short tandem repeats bind transcription factors to tune eukaryotic gene expression
Horton, C. A., Alexandari, A. M., Hayes, M. G. B., Marklund, E., Schaepe, J. M., Aditham, A. K., Shah, N., Suzuki, P., Shrikumar, A., Afek, A., et al. (2023) · Science · doi:10.1126/science.add1250 · Cited by 207
Self-citationSequence-based modeling of three-dimensional genome architecture from kilobase to chromosome scale
Zhou, J. (2022) · Nature Genetics · doi:10.1038/s41588-022-01065-4 · Cited by 177
“Decoders at lower levels receive input from the corresponding level of sequence representations, the interaction matrix prediction from one-level above and a 2D pairwise distance encoding matrix as an auxiliary input.”
BackgroundDecoding gene regulation in the fly brain
Janssens, J., Aibar, S., Taskiran, I. I., Ismail, J. N., Gomez, A. E., Aughey, G., Spanier, K. I., De Rop, F. V., González-Blas, C. B., Dionne, M., et al. (2022) · Nature · doi:10.1038/s41586-021-04262-z · Cited by 170
Decoding gene regulation in the fly brain
Janssens, J., Aibar, S., Taskiran, I. I., Ismail, J. N., Spanier, K. I., Bravo González-Blas, C., Quan, X. J., Papasokrati, D., Hulselmans, G., Makhzami, S., et al. (2021) · doi:10.1101/2021.08.11.454937 · Cited by 170
Cell-type-directed design of synthetic enhancers
Taskiran, I. I., Spanier, K. I., Dickmänken, H., Kempynck, N., Pančíková, A., Ekşi, E. C., Hulselmans, G., Ismail, J. N., Theunis, K., Vandepoel, R., et al. (2024) · Nature · doi:10.1038/s41586-023-06936-2 · Cited by 175
“Here we show that deep learning models 2–6 , can be used to efficiently design synthetic, cell-type-specific enhancers, starting from random sequences, and that this optimization process allows detailed tracing of enhancer features at single-nucleotide resolution.”
BackgroundInfluentialUyehara, C. M., & Apostolou, E. (2023) · Cell Reports · doi:10.1016/j.celrep.2023.112068 · Cited by 143
Enhancer grammar in development, evolution, and disease: dependencies and interplay
Jindal, G. A., & Farley, E. K. (2021) · Developmental Cell · doi:10.1016/j.devcel.2021.02.016 · Cited by 152
Machine-guided design of cell-type-targeting cis-regulatory elements
Gosai, S. J., Castro, R. I., Fuentes, N., Butts, J. C., Mouri, K., Alasoadura, M., Kales, S., Nguyen, T. T. L., Noche, R. R., Rao, A. S., et al. (2024) · Nature · doi:10.1038/s41586-024-08070-z · Cited by 142
“Over the course of this study, several advances in DNA modelling have been reported by other groups that would probably yield such improvements 14,25,36,81,82 , but were not tested here.”
Pampari, A., Shcherbina, A., Kvon, E. Z., Kosicki, M., Nair, S., Kundu, S., Kathiria, A. S., Risca, V. I., Kuningas, K., Alasoo, K., et al. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.12.25.630221 · Cited by 183
Self-citationSasse, A., Ng, B., Spiro, A. E., Tasaki, S., Bennett, D. A., Gaiteri, C., De Jager, P. L., Chikina, M., & Mostafavi, S. (2023) · Nature Genetics · doi:10.1038/s41588-023-01524-6 · Cited by 139
“State-of-the-art models can reproduce experimental measurements with a high degree of accuracy and enable mechanistic insights through their learned DNA features 1,2,15 .”
AI applications in functional genomics
Caudai, C., Galizia, A., Geraci, F., Pera, L. L., Morea, V., Salerno, E., Via, A., & Colombo, T. (2021) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2021.10.009 · Cited by 155
“More recently, Avsec et al. [155] introduced a DL approach to unravel the influence of motif spacing between neighbour transcription-factor binding sites on transcription factor cooperativity.”
BackgroundKarollus, A., Mauermeier, T., & Gagneur, J. (2023) · Genome Biology · doi:10.1186/s13059-023-02899-9 · Cited by 128
“A recent study, for example, used a deep model to analyze the impact of helical periodicity on TF binding [30].”
BackgroundSelf-citationSong, Z., Huang, D., Song, B., Chen, K., Song, Y., Liu, G., Su, J., de Magalhães, J. P., Rigden, D. J., & Meng, J. (2021) · Nature Communications · doi:10.1038/s41467-021-24313-3 · Cited by 142
scBasset: sequence-based modeling of single-cell ATAC-seq using convolutional neural networks
Yuan, H., & Kelley, D. R. (2022) · Nature Methods · doi:10.1038/s41592-022-01562-8 · Cited by 133
“Methods such as TF-MoDISco could be applied to scBasset ISM scores for de novo motif discovery 14,34 .”
MethodMachine learning for perturbational single-cell omics
Ji, Y., Lotfollahi, M., Wolf, F. A., & Theis, F. J. (2021) · Cell Systems · doi:10.1016/j.cels.2021.05.016 · Cited by 137
Integrating multi-omics data with EHR for precision medicine using advanced artificial intelligence
Tong, L., Shi, W., Isgut, M., Zhong, Y., Lais, P., Gloster, L., Sun, J., Swain, A., Giuste, F., & Wang, M. D. (2024) · IEEE Reviews in Biomedical Engineering · doi:10.1109/rbme.2023.3324264 · Cited by 130
“Avsec et al. [158] proposed a Base Pair Network to reveal regulatory code by accurately predicting transcription factor binding from DNA sequences.”
BackgroundPersonal transcriptome variation is poorly explained by current genomic deep learning models
Huang, C., Shuai, R. W., Baokar, P., Chung, R., Rastogi, R., Kathail, P., & Ioannidis, N. M. (2023) · Nature Genetics · doi:10.1038/s41588-023-01574-w · Cited by 127
“These genomic deep learning models are trained using genome-wide data from a variety of cell types and cellular contexts and have been shown to learn biologically relevant regulatory motifs within the input DNA sequence [8, 9].”
Tan, J., Shenker-Tauris, N., Rodriguez-Hernaez, J., Wang, E., Sakellaropoulos, T., Boccalatte, F. E., Thandapani, P., Skok, J. A., Aifantis, I., Fenyö, D., et al. (2023) · Nature Biotechnology · doi:10.1038/s41587-022-01612-8 · Cited by 123
“Building on C.Origami’s model architecture, we first developed two fast approaches”
Mammalian evolution of human cis-regulatory elements and transcription factor binding sites
Andrews, G., Fan, K., Pratt, H. E., Phalke, N. N., Karlsson, E. K., Lindblad‐Toh, K., Gazal, S., Moore, J. E., Weng, Z., Andrews, G., et al. (2023) · Science · doi:10.1126/science.abn7930 · Cited by 110
Ovek, D., Rauluševičiūtė, I., Aronsen, D. R., Blanc‐Mathieu, R., Bonthuis, I., Beukelaer, H. D., Ferenc, K., Jegou, A., Kumar, V., Lemma, R. B., et al. (2025) · Nucleic Acids Research · doi:10.1093/nar/gkaf1209 · Cited by 118
“We trained a specific BPNet [ 25 ] model for each ChIP-seq dataset to predict the ChIP-seq genomic tracks at base-pair resolution from input DNA sequences (see Supplementary Fig.”
InfluentialSelf-citationChen, V., Yang, M., Cui, W., Kim, J. S., Talwalkar, A., & Ma, J. (2024) · Nature Methods · doi:10.1038/s41592-024-02359-7 · Cited by 144
Cross-species analysis of enhancer logic using deep learning
Minnoye, L., Taskiran, I. I., Mauduit, D., Fazio, M., Van Aerschot, L., Hulselmans, G., Christiaens, V., Makhzami, S., Seltenhammer, M., Karras, P., et al. (2020) · Genome Research · doi:10.1101/gr.260844.120 · Cited by 116
“2018), TF binding sites (Wang et al. 2018; Avsec et al. 2020), DNAmethylation (Angermueller et al.”
BackgroundMethodAmeen, M., Sundaram, L., Shen, M., Banerjee, A., Kundu, S., Nair, S., Shcherbina, A., Gu, M., Wilson, K. D., Varadarajan, A., et al. (2022) · Cell · doi:10.1016/j.cell.2022.11.028 · Cited by 113
Self-citationLinder, J., Bogard, N., Rosenberg, A. B., & Seelig, G. (2020) · Cell Systems · doi:10.1016/j.cels.2020.05.007 · Cited by 107
Hold out the genome: a roadmap to solving the cis-regulatory code
de Boer, C. G., & Taipale, J. (2024) · Nature · doi:10.1038/s41586-023-06661-w · Cited by 107
Sequence basis of transcription initiation in the human genome
Dudnyk, K., Cai, D., Shi, C., Xu, J., & Zhou, J. (2024) · Science · doi:10.1126/science.adj0116 · Cited by 115
Seven myths of how transcription factors read the cis-regulatory code
Zeitlinger, J. (2020) · Current Opinion in Systems Biology · doi:10.1016/j.coisb.2020.08.002 · Cited by 105
Self-citationAlQuraishi, M., & Sorger, P. K. (2021) · Nature Methods · doi:10.1038/s41592-021-01283-4 · Cited by 96
Transformers and genome language models
Consens, M. E., Dufault, C., Wainberg, M., Forster, D., Karimzadeh, M., Goodarzi, H., Theis, F. J., Moses, A. M., & Wang, B. (2025) · Nature Machine Intelligence · doi:10.1038/s42256-025-01007-9 · Cited by 170
Deep flanking sequence engineering for efficient promoter design using DeepSEED
Zhang, P., Wang, H., Xu, H., Wei, L., Liu, L., Hu, Z., & Wang, X. (2023) · Nature Communications · doi:10.1038/s41467-023-41899-y · Cited by 94
Deciphering the impact of genomic variation on function
Writing group (ordered by contribution), Engreitz, J., Lawson, H. A., Singh, H., Starita, L. M., Hon, G. C., Carter, H. K., Sahni, N., Reddy, T. E., Lin, X., et al. (2024) · Nature · doi:10.1038/s41586-024-07510-0 · Cited by 105
“…that generalize across contexts—for example, to link genetic variants to effects on transcription factor binding and chromatin accessibility 57,61–64 , connect regulatory elements to their target genes 64,85,86 , or identify causal genes and cell types enriched for heritability for complex…”
Self-citationGenome-wide mapping of somatic mutation rates uncovers drivers of cancer
Sherman, M. A., Yaari, A. U., Priebe, O., Dietlein, F., Loh, P.-R., & Berger, B. (2022) · Nature Biotechnology · doi:10.1038/s41587-022-01353-8 · Cited by 80
Wang, S. K., Nair, S., Li, R., Kraft, K., Pampari, A., Patel, A., Kang, J. B., Luong, C., Kundaje, A., & Chang, H. Y. (2022) · bioRxiv · doi:10.1016/j.xgen.2022.100164 · Cited by 101
“As a final method to prioritize SNPs in our dataset, we therefore trained CNNs derived from the BPNet architecture on scATAC-seq profiles for each of the 13 retinal cell types (Figure 6A, S9A, and S9B).(12) At each SNP region, we compared the projected per-base change in chromatin accessibility between reference and a…”
BackgroundMethodSelf-citationDecoding disease: from genomes to networks to phenotypes
Wong, A. K., Sealfon, R. S. G., Theesfeld, C. L., & Troyanskaya, O. G. (2021) · Nature Reviews Genetics · doi:10.1038/s41576-021-00389-x · Cited by 90
Deep learning the cis-regulatory code for gene expression in selected model plants
Peleke, F. F., Zumkeller, S. M., Gültaş, M., Schmitt, A. O., & Szymański, J. (2024) · Nature Communications · doi:10.1038/s41467-024-47744-0 · Cited by 98
“In contrast, to the 2D convolutional and max pooling layers used by Washburn and colleagues 18 , we used 1D layers asit has been the choicein morerecent studies inferring gene expression or protein interaction from DNA sequence using deep learning, e.g. the Enformer or BPNet architecture 17,21 .”
Deep learning in CRISPR-Cas systems: a review of recent studies
Lee, M. (2023) · Frontiers in Bioengineering and Biotechnology · doi:10.3389/fbioe.2023.1226182 · Cited by 89
“Similar predictive capabilities were showcased by Avsec et al. (2021) through their BPNet model, enabling the discovery of soft syntax rules for cooperative transcription factor binding interactions. Lastly, Muller et al. (2020) utilized deep learning to gauge public opinions on CRISPR-Cas9, providing a compelling per…”
BackgroundInfluentialPosition-dependent function of human sequence-specific transcription factors
Duttke, S. H. C., Guzman, C., Chang, M. W., Santos, N. P. D., McDonald, B. R., Xie, J., Carlin, A. F., Heinz, S., & Benner, C. W. (2024) · Nature · doi:10.1038/s41586-024-07662-z · Cited by 89
“Enrichment patterns for several TF binding sites exhibited an approximately 10 bp periodicity, suggesting that the rotational position of TFs relative to the TSS affects transcription initiation 5,14,35 (Extended Data Fig.”
Current challenges in understanding the role of enhancers in disease
Zaugg, J. B., Sahlén, P., Andersson, R., Alberich-Jordà, M., de Laat, W. L., Deplancke, B., Ferrer, J. P., Mandrup, S., Natoli, G., Plewczyński, D., et al. (2022) · Nature Structural & Molecular Biology · doi:10.1038/s41594-022-00896-3 · Cited by 86
Transcriptional enhancers and their communication with gene promoters
Ray-Jones, H., & Spivakov, M. (2021) · Cellular and Molecular Life Sciences · doi:10.1007/s00018-021-03903-w · Cited by 81
“Additionally, a deep learning algorithm applied to binding of pluripotency TFs at base resolution in mouse embryonic stem cells (ESCs) suggested a “soft motif syntax” whereby TFs help each other to bind directly to the enhancer with a lenient distance-dependency between binding sites [98].”
BackgroundKoo, P. K., Majdandzic, A., Ploenzke, M., Anand, P., & Paul, S. B. (2021) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1008925 · Cited by 78
“Dilated convolutions combined with skipped connections have been previously employed in various settings for regulatory genomics [16, 17, 41].”
BackgroundMethodScalable Functional Assays for the Interpretation of Human Genetic Variation
Tabet, D. R., Parikh, V. N., Mali, P., Roth, F. P., & Claussnitzer, M. (2022) · Annual Review of Genetics · doi:10.1146/annurev-genet-072920-032107 · Cited by 84
“Extensive genome-scale measurements of expression, chromatin accessibility, and epigenetic marks have fueled the development of models to predict gene regulatory variant effects from DNA sequences (13, 14, 29, 40, 103, 204).”
BackgroundGeorgakopoulos-Soares, I., Deng, C., Agarwal, V., Chan, C. S. Y., Zhao, J., Inoue, F., & Ahituv, N. (2023) · Nature Communications · doi:10.1038/s41467-023-37960-5 · Cited by 82
“In addition, in previous work, we identi fi ed widespread transcriptional strand asymmetries in the orientation of TFBSs across human transcribed regions 10 , a research area in which there have been a few studies to date 11 – 15 .”
The dynamic, combinatorial cis-regulatory lexicon of epidermal differentiation
Kim, D. S., Risca, V. I., Reynolds, D. L., Chappell, J., Rubin, A. J., Jung, N., Donohue, L. K. H., Lopez-Pajares, V., Kathiria, A. S., Shi, M., et al. (2021) · Nature Genetics · doi:10.1038/s41588-021-00947-3 · Cited by 80
Self-citationInterpreting non-coding disease-associated human variants using single-cell epigenomics
Gaulton, K. J., Preißl, S., & Ren, B. (2023) · Nature Reviews Genetics · doi:10.1038/s41576-023-00598-6 · Cited by 81
“A neural network, BPNet 133 , was used to model cell type-specific cCRE activity in the developing cerebral cortex at base-pair resolution, which was then applied to predict variants with functional effects in each cortex cell type.”
MethodBrennan, K., Weilert, M., Krueger, S., Pampari, A., Liu, H., Yang, A. W. H., Morrison, J. A., Hughes, T. R., Rushlow, C. A., Kundaje, A. B., et al. (2023) · Developmental Cell · doi:10.1016/j.devcel.2023.07.007 · Cited by 79
“To learn DNA sequence rules at the highest possible resolution in the early Drosophila embryo, we used ChIP-nexus, a chromatin immunoprecipitation technique that maps genome-wide TF binding footprints at base-resolution by virtue of a strandspecific exonuclease,(58) and employed a CNN called BPNet, which directly pred…”
Self-citationPredicting gene expression from DNA sequence using deep learning models
Barbadilla-Martínez, L., Klaassen, N. H. M., van Steensel, B., & de Ridder, J. (2025) · Nature Reviews Genetics · doi:10.1038/s41576-025-00841-2 · Cited by 89
Deep learning in next-generation sequencing
Schmidt, B., & Hildebrandt, A. (2020) · Drug Discovery Today · doi:10.1016/j.drudis.2020.10.002 · Cited by 80
“A recent example in the context of NGS is the design of an interpretation tool that learns predictive motif representations for cooperative transcription factor binding interactions from ChIP-nexus sequencing data using DeepLift [52].”
MethodZEB2 Shapes the Epigenetic Landscape of Atherosclerosis
Cheng, P. P., Wirka, R. C., Clarke, S. L., Zhao, Q., Kundu, R. K., Nguyen, T., Nair, S., Sharma, D., Kim, H.-J., Shi, H., et al. (2022) · Circulation · doi:10.1161/circulationaha.121.057789 · Cited by 70
Self-citationSelective gene expression maintains human tRNA anticodon pools during differentiation
Gao, L., Behrens, A., Rodschinka, G., Forcelloni, S., Wani, S., Strasser, K., & Nedialkova, D. D. (2024) · Nature Cell Biology · doi:10.1038/s41556-023-01317-3 · Cited by 78
“Conceptually, the architecture is based on that described for BPNet 56 with minor adjustments to the size of the receptive field of the network and the output (Extended Data Fig.”
Chromatin interaction–aware gene regulatory modeling with graph attention networksPreprint
Karbalayghareh, A., Şahin, M., & Leslie, C. S. (2022) · Genome Research · doi:10.1101/gr.275870.121 · Cited by 76
“Dilated CNNs have been used previously to increase the receptive field of CNN layers in deep learning models that predict epigenomic data from DNA sequence, such as Basenji [7], BPNet [14], and Xpresso [15].”
MethodSynthetic regulatory genomics uncovers enhancer context dependence at the Sox2 locus
Brosh, R., Coelho, C., Ribeiro-dos-Santos, A. M., Ellis, G., Hogan, M. S., Ashe, H. J., Somogyi, N., Ordóñez, R., Luther, R. D., Huang, E., et al. (2023) · Molecular Cell · doi:10.1016/j.molcel.2023.02.027 · Cited by 75
High-throughput PRIME-editing screens identify functional DNA variants in the human genome
Ren, X., Yang, H., Nierenberg, J. L., Sun, Y., Chen, J., Beaman, C., Pham, T., Nobuhara, M., Takagi, M. A., Narayan, V., et al. (2023) · Molecular Cell · doi:10.1016/j.molcel.2023.11.021 · Cited by 70
Decoding gene regulation in the mouse embryo using single-cell multi-omicsPreprint
Argelaguet, R., Lohoff, T., Li, J., Nakhuda, A., Drage, D., Krueger, F., Velten, L., Clark, S. J., & Reik, W. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.06.15.496239 · Cited by 64
“This stands in contrast with other approaches that have been proposed to predict TF binding from multi-omics data, which employ supervised models that require labelled training data from ChIP-seq experiments (Avsec et al., 2021; Karimzadeh and Hoffman, 2019).”
ResultApplications of deep learning in understanding gene regulation
Li, Z., Gao, E., Zhou, J., Han, W., Xu, X., & Gao, X. (2023) · Cell Reports Methods · doi:10.1016/j.crmeth.2022.100384 · Cited by 73
“48 Model-based methods Instead of making post hoc interpretations of the model, it is better to consider interpretability during the model’s development.”
MethodDeep learning and CRISPR-Cas13d ortholog discovery for optimized RNA targeting
Wei, J., Lotfy, P., Faizi, K., Baungaard, S., Gibson, E., Wang, E., Slabodkin, H., Kinnaman, E., Chandrasekaran, S. S., Kitano, H. C., et al. (2023) · Cell Systems · doi:10.1016/j.cels.2023.11.006 · Cited by 71
Predicting mean ribosome load for 5’UTR of any length using deep learning
Karollus, A., Avsec, Ž., & Gagneur, J. (2021) · PLOS Computational Biology · doi:10.1371/journal.pcbi.1008982 · Cited by 57
Self-citationTu, X., Ren, S., Shen, W., Li, J., Li, Y., Li, C., Li, Y., Zong, Z., Xie, W., Grierson, D. S., et al. (2022) · Nature Communications · doi:10.1038/s41467-022-35438-4 · Cited by 67
Evaluating deep learning for predicting epigenomic profiles
Toneyan, S., Tang, Z., & Koo, P. K. (2022) · Nature Machine Intelligence · doi:10.1038/s42256-022-00570-9 · Cited by 64
Single-cell chromatin accessibility reveals malignant regulatory programs in primary human cancers
Sundaram, L., Kumar, A., Zatzman, M., Salcedo, A., Ravindra, N. G., Shams, S., Louie, B. H., Bagdatli, S. T., Myers, M. A., Sarmashghi, S., et al. (2024) · Science · doi:10.1126/science.adk9217 · Cited by 66
“…incancers,wetrainedaconvolutionalneural networktolearnamappingfrom1364 – base pair (bp) DNA sequence windows around scATAC-seq peaks and background regions to the pseudobulk peak probability, similar to previous methods ( 9 , 30 – 32 ) (Fig.”
MethodInterpretable detection of novel human viruses from genome sequencing data
Bartoszewicz, J. M., Seidel, A., & Renard, B. Y. (2021) · NAR Genomics and Bioinformatics · doi:10.1093/nargab/lqab004 · Cited by 67
(Alternative) transcription start sites as regulators of RNA processing
Alfonso-Gonzalez, C., & Hilgers, V. (2024) · Trends in Cell Biology · doi:10.1016/j.tcb.2024.02.010 · Cited by 67
Hoffman, G. E., Bendl, J. M., Girdhar, K., Schadt, E. E., & Roussos, P. (2019) · Nucleic Acids Research · doi:10.1093/nar/gkz808 · Cited by 63
“Moreover, there is likely meaningful epigenetic variation outside of strictly defined peak regions and, indeed, there is much interest in performing basepair-level predictions rather than summarizing the epigenetic signal for each peak (28,78).”
BackgroundJanizek, J. D., Dincer, A. B., Çelik, S., Chen, H., Chen, W., Naxerova, K., & Lee, S. (2023) · Nature Biomedical Engineering · doi:10.1038/s41551-023-01034-0 · Cited by 66
An integrated view of the structure and function of the human 4D nucleome
Dekker, J., Oksuz, B. A., Zhang, Y., Wang, Y., Minsk, M. K., Kuang, S., Yang, L., Gibcus, J. H., Krietenstein, N., Rando, O. J., et al. (2025) · Nature · doi:10.1038/s41586-025-09890-3 · Cited by 64
“This can be done using explainable artificial intelligence techniques, such as importance scores 128 and in silico mutagenesis 129,130 .”
Evaluating the representational power of pre-trained DNA language models for regulatory genomics
Tang, Z., Somia, N., Yu, Y., & Koo, P. K. (2025) · Genome biology · doi:10.1186/s13059-025-03674-8 · Cited by 71
“By contrast, supervised deep learning models trained on large troves of functional genomics data in a multitask setting can learn discriminative features related to cis -regulatory mechanisms in the non-coding genome [70, 92–97].”
Deciphering the impact of genetic variation on human polyadenylation using APARENT2
Linder, J., Koplik, S. E., Kundaje, A., & Seelig, G. (2022) · Genome Biology · doi:10.1186/s13059-022-02799-4 · Cited by 49
Self-citationRay, A., Das, J., & Wenzel, S. E. (2022) · Cell Reports Medicine · doi:10.1016/j.xcrm.2022.100857 · Cited by 64
“Further, while deep learning approaches have traditionally been hard to interpret, recent techniques have creatively used techniques like backpropagation to make these interpretable and amenable to inference beyond prediction.(118) These methods can integrate multi-omic datasets to uncover signatures of asthma molecul…”
MethodSong, B., Buckler, E. S., Wang, H., Wu, Y., Rees, E. R., Kellogg, E. A., Gates, D. J., Khaipho-Burch, M., Bradbury, P. J., Ross‐Ibarra, J., et al. (2021) · Genome Research · doi:10.1101/gr.266528.120 · Cited by 57
“TFs often work cooperatively and the sequence contexts of TF recognition sites or combinatorial recognition of cis-elements are key for TF binding specificity (Tu et al. 2020; Gerstein et al. 2012; O’Malley et al. 2016; Dror et al. 2015; Levo et al. 2015; Avsec et al. 2021).”
BackgroundKing, H. W., Wells, K. L., Shipony, Z., Kathiria, A. S., Wagar, L. E., Lareau, C. A., Orban, N. T., Capasso, R., Davis, M. M., Steinmetz, L. M., et al. (2021) · Science Immunology · doi:10.1126/sciimmunol.abh3768 · Cited by 55
“New advances in neural network–derived methods may prove useful to quantitatively model effects on gene expression in cell type–resolved chromatin accessibility maps ( 17 , 54 ).”
Multiplexed single-cell characterization of alternative polyadenylation regulators
Kowalski, M. H., Wessels, H., Linder, J., Dalgarno, C., Mascio, I., Choudhary, S., Hartman, A., Hao, Y., Kundaje, A. B., & Satija, R. (2024) · Cell · doi:10.1016/j.cell.2024.06.005 · Cited by 56
Self-citationA community effort to optimize sequence-based deep learning models of gene regulation
Rafi, A. M., Nogina, D., Penzar, D., Lee, D., Lee, D., KIM, N. S., Kim, S., Kim, D., Shin, Y., Kwak, I., et al. (2024) · Nature Biotechnology · doi:10.1038/s41587-024-02414-w · Cited by 63
Lu, A. X., Lu, A. X., Pritišanac, I., Zarin, T., Forman-Kay, J. D., & Moses, A. M. (2022) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1010238 · Cited by 56
Rewriting regulatory DNA to dissect and reprogram gene expression
Martyn, G. E., Montgomery, M. T., Jones, H., Guo, K., Doughty, B. R., Linder, J., Bisht, D., Xia, F., Cai, X. S., Chen, Z., et al. (2023) · Cell · doi:10.1016/j.cell.2025.03.034 · Cited by 59
Self-citationButz, S., Schmolka, N., Karemaker, I. D., Villaseñor, R., Schwarz, I., Domcke, S., Uijttewaal, E. C. H., Jude, J., Lienert, F., Krebs, A. R., et al. (2022) · Nature Genetics · doi:10.1038/s41588-022-01210-z · Cited by 44
“The differential DNA methylation states at the ectopic Airn ICR were stably maintained after prolonged cultivation https://doi.org/10.1038/s41588-022-01210-z lack of maintenance, in agreement with an in silico evaluation of ZFP57 binding to wild-type and shuffled Airn ICR sequence using BPNet 28 (Extended Data Fig.”
InfluentialFactorbook: an updated catalog of transcription factor motifs and candidate regulatory motif sites
Pratt, H., Andrews, G., Phalke, N. N., Huey, J. D., Purcaro, M. J., van der Velde, A., Moore, J. E., & Weng, Z. (2021) · Nucleic Acids Research · doi:10.1093/nar/gkab1039 · Cited by 52
Correcting gradient-based interpretations of deep neural networks for genomics
Majdandzic, A., Rajesh, C., & Koo, P. K. (2023) · Genome biology · doi:10.1186/s13059-023-02956-3 · Cited by 56
“To gain insights into the features learned by DNNs, post-hoc attribution methods provide an importance score for each nucleotide in a given sequence; they often reveal biologically meaningful patterns, such as transcription factor binding motifs that are essential for gene regulation [4, 5].”
BackgroundInterpretation of allele-specific chromatin accessibility using cell state–aware deep learning
Atak, Z. K., Taskiran, I. I., Demeulemeester, J., Flerin, C., Mauduit, D., Minnoye, L., Hulselmans, G., Christiaens, V., Ghanem, G.-E., Wouters, J., et al. (2021) · Genome Research · doi:10.1101/gr.260851.120 · Cited by 48
“However, thesemodels require correct training and rigorous validation and are typically trained either for a single TF (Alipanahi et al. 2015; Lee 2016; Quang and Xie 2019; Avsec et al. 2021) or for hundreds of epigenomic features (Zhou and Troyanskaya 2015; Kelley et al. 2016).”
BackgroundMachine learning applications for therapeutic tasks with genomics data
Huang, K., Xiao, C., Glass, L. M., Critchlow, C. W., Gibson, G. C., & Sun, J. (2021) · Patterns · doi:10.1016/j.patter.2021.100328 · Cited by 46
“Recently, Avsec et al.61 have shown the benefits of the base-resolution CNN model in TF binding prediction.”
Background
Metadata
- Authors
- Avsec, Ž., Weilert, M., Shrikumar, A., Krueger, S., Alexandari, A., Dalal, K., Fropf, R., McAnany, C., Gagneur, J., Kundaje, A., & Zeitlinger, J.
- Affiliations
- Google DeepMind (United Kingdom) (GB)
- Stowers Institute for Medical Research (US)
- Stanford University (US)
- Technical University of Munich (DE)
- Corresponding author
- Kundaje, Anshul
- Zeitlinger, Julia
- ORCID
- Venue
- Nature Genetics
- Published
- 2021-02-18
- Volume
- 53
- Issue
- 3
- Pages
- 354-366
- ISSN
- 1061-4036 (ISSN-L)
- 1546-1718
- PMC
- PMC8812996
- OpenAlex
- W3130819465
- Cited by
- 610
- References
- 163
- Citations by year
- 2026: 118 · 2025: 108 · 2024: 124 · 2023: 102 · 2022: 90 · 2021: 49
- FWCI
- 29.88
- Type
- journal article
- Language
- en
- Open access
- green
- doi.org
- Licence
- Locations
- Nature Genetics (publishedVersion)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nc)
- PubMed (publishedVersion)
- PubMed Central (acceptedVersion)
- Funders
- U.S. Department of Health & Human Services | NIH | National Human Genome Research Institute (1R01HG010211, 1U01HG009431, 1R01HG009674)
- Howard Hughes Medical Institute (International Student Research Fellowship)
- U.S. Department of Health & Human Services | NIH | National Institute of General Medical Sciences (1DP2GM123485)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Chromatin Dynamics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene Regulatory Network Analysis
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Epigenetics and DNA Methylation
- Keywords
- transcription factor binding, pluripotency transcription factors, Nanog, helical periodicity, explainable deep learning
- MeSH terms
- Mouse Embryonic Stem Cells, Nanog Homeobox Protein, Deep Learning, Animals, Binding Sites, Transcription Factors, Reproducibility of Results, Neural Networks, Computer, Computational Biology, Chromatin Immunoprecipitation, Octamer Transcription Factor-3, Mice, SOXB1 Transcription Factors, Nucleotide Motifs, Clustered Regularly Interspaced Short Palindromic Repeats
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
