Kelly Cochran

All Publications

• GENCODE 2025: reference gene annotation for human and mouse. Nucleic acids research Mudge, J. M., Carbonell-Sala, S., Diekhans, M., Martinez, J. G., Hunt, T., Jungreis, I., Loveland, J. E., Arnan, C., Barnes, I., Bennett, R., Berry, A., Bignell, A., Cerdán-Vélez, D., Cochran, K., Cortés, L. T., Davidson, C., Donaldson, S., Dursun, C., Fatima, R., Hardy, M., Hebbar, P., Hollis, Z., James, B. T., Jiang, Y., Johnson, R., Kaur, G., Kay, M., Mangan, R. J., Maquedano, M., Gómez, L. M., Mathlouthi, N., Merritt, R., Ni, P., Palumbo, E., Perteghella, T., Pozo, F., Raj, S., Sisu, C., Steed, E., Sumathipala, D., Suner, M. M., Uszczynska-Ratajczak, B., Wass, E., Yang, Y. T., Zhang, D., Finn, R. D., Gerstein, M., Guigó, R., Hubbard, T. J., Kellis, M., Kundaje, A., Paten, B., Tress, M. L., Birney, E., Martin, F. J., Frankish, A. 2024

  Abstract

  GENCODE produces comprehensive reference gene annotation for human and mouse. Entering its twentieth year, the project remains highly active as new technologies and methodologies allow us to catalog the genome at ever-increasing granularity. In particular, long-read transcriptome sequencing enables us to identify large numbers of missing transcripts and to substantially improve existing models, and our long non-coding RNA catalogs have undergone a dramatic expansion and reconfiguration as a result. Meanwhile, we are incorporating data from state-of-the-art proteomics and Ribo-seq experiments to fine-tune our annotation of translated sequences, while further insights into function can be gained from multi-genome alignments that grow richer as more species' genomes are sequenced. Such methodologies are combined into a fully integrated annotation workflow. However, the increasing complexity of our resources can present usability challenges, and we are resolving these with the creation of filtered genesets such as MANE Select and GENCODE Primary. The next challenge is to propagate annotations throughout multiple human and mouse genomes, as we enter the pangenome era. Our resources are freely available at our web portal www.gencodegenes.org, and via the Ensembl and UCSC genome browsers.

  View details for DOI 10.1093/nar/gkae1078

  View details for PubMedID 39565199

• GENCODE: massively expanding the lncRNA catalog through capture long-read RNA sequencing. bioRxiv : the preprint server for biology Kaur, G., Perteghella, T., Carbonell-Sala, S., Gonzalez-Martinez, J., Hunt, T., Mądry, T., Jungreis, I., Arnan, C., Lagarde, J., Borsari, B., Sisu, C., Jiang, Y., Bennett, R., Berry, A., Cerdán-Vélez, D., Cochran, K., Vara, C., Davidson, C., Donaldson, S., Dursun, C., González-López, S., Gopal Das, S., Hardy, M., Hollis, Z., Kay, M., Montañés, J. C., Ni, P., Nurtdinov, R., Palumbo, E., Pulido-Quetglas, C., Suner, M. M., Yu, X., Zhang, D., Loveland, J. E., Albà, M. M., Diekhans, M., Tanzer, A., Mudge, J. M., Flicek, P., Martin, F. J., Gerstein, M., Kellis, M., Kundaje, A., Paten, B., Tress, M. L., Johnson, R., Uszczynska-Ratajczak, B., Frankish, A., Guigó, R. 2024

  Abstract

  Accurate and complete gene annotations are indispensable for understanding how genome sequences encode biological functions. For twenty years, the GENCODE consortium has developed reference annotations for the human and mouse genomes, becoming a foundation for biomedical and genomics communities worldwide. Nevertheless, collections of important yet poorly-understood gene classes like long non-coding RNAs (lncRNAs) remain incomplete and scattered across multiple, uncoordinated catalogs, slowing down progress in the field. To address these issues, GENCODE has undertaken the most comprehensive lncRNAs annotation effort to date. This is founded on the manual annotation of full-length targeted long-read sequencing, on matched embryonic and adult tissues, of orthologous regions in human and mouse. Altogether 17,931 novel human genes (140,268 novel transcripts) and 22,784 novel mouse genes (136,169 novel transcripts) have been added to the GENCODE catalog representing a 2-fold and 6-fold increase in transcripts, respectively - the greatest increase since the sequencing of the human genome. Novel gene annotations display evolutionary constraints, have well-formed promoter regions, and link to phenotype-associated genetic variants. They greatly enhance the functional interpretability of the human genome, as they help explain millions of previously-mapped "orphan" omics measurements corresponding to transcription start sites, chromatin modifications and transcription factor binding sites. Crucially, our targeted design assigned human-mouse orthologs at a rate beyond previous studies, tripling the number of human disease-associated lncRNAs with mouse orthologs. The expanded and enhanced GENCODE lncRNA annotations mark a critical step towards deciphering the human and mouse genomes.

  View details for DOI 10.1101/2024.10.29.620654

  View details for PubMedID 39554180

  View details for PubMedCentralID PMC11565817

• Dissecting the cis-regulatory syntax of transcription initiation with deep learning. bioRxiv : the preprint server for biology Cochran, K., Yin, M., Mantripragada, A., Schreiber, J., Marinov, G. K., Kundaje, A. 2024

  Abstract

  Despite extensive characterization of mammalian Pol II transcription, the DNA sequence determinants of transcription initiation at a third of human promoters and most enhancers remain poorly understood. Hence, we trained and interpreted a neural network called ProCapNet that accurately models base-resolution initiation profiles from PRO-cap experiments using local DNA sequence. ProCapNet learns sequence motifs with distinct effects on initiation rates and TSS positioning and uncovers context-specific cryptic initiator elements intertwined within other TF motifs. ProCapNet annotates predictive motifs in nearly all actively transcribed regulatory elements across multiple cell-lines, revealing a shared cis-regulatory logic across promoters and enhancers mediated by a highly epistatic sequence syntax of cooperative and competitive motif interactions. ProCapNet models of RAMPAGE profiles measuring steady-state RNA abundance at TSSs distill initiation signals on par with models trained directly on PRO-cap profiles. ProCapNet learns a largely cell-type-agnostic cis-regulatory code of initiation complementing sequence drivers of cell-type-specific chromatin state critical for accurate prediction of cell-type-specific transcription initiation.

  View details for DOI 10.1101/2024.05.28.596138

  View details for PubMedID 38853896

• Rewriting regulatory DNA to dissect and reprogram gene expression. bioRxiv : the preprint server for biology Martyn, G. E., Montgomery, M. T., Jones, H., Guo, K., Doughty, B. R., Linder, J., Chen, Z., Cochran, K., Lawrence, K. A., Munson, G., Pampari, A., Fulco, C. P., Kelley, D. R., Lander, E. S., Kundaje, A., Engreitz, J. M. 2023

  Abstract

  Regulatory DNA sequences within enhancers and promoters bind transcription factors to encode cell type-specific patterns of gene expression. However, the regulatory effects and programmability of such DNA sequences remain difficult to map or predict because we have lacked scalable methods to precisely edit regulatory DNA and quantify the effects in an endogenous genomic context. Here we present an approach to measure the quantitative effects of hundreds of designed DNA sequence variants on gene expression, by combining pooled CRISPR prime editing with RNA fluorescence in situ hybridization and cell sorting (Variant-FlowFISH). We apply this method to mutagenize and rewrite regulatory DNA sequences in an enhancer and the promoter of PPIF in two immune cell lines. Of 672 variant-cell type pairs, we identify 497 that affect PPIF expression. These variants appear to act through a variety of mechanisms including disruption or optimization of existing transcription factor binding sites, as well as creation of de novo sites. Disrupting a single endogenous transcription factor binding site often led to large changes in expression (up to -40% in the enhancer, and -50% in the promoter). The same variant often had different effects across cell types and states, demonstrating a highly tunable regulatory landscape. We use these data to benchmark performance of sequence-based predictive models of gene regulation, and find that certain types of variants are not accurately predicted by existing models. Finally, we computationally design 185 small sequence variants (≤10 bp) and optimize them for specific effects on expression in silico. 84% of these rationally designed edits showed the intended direction of effect, and some had dramatic effects on expression (-100% to +202%). Variant-FlowFISH thus provides a powerful tool to map the effects of variants and transcription factor binding sites on gene expression, test and improve computational models of gene regulation, and reprogram regulatory DNA.

  View details for DOI 10.1101/2023.12.20.572268

  View details for PubMedID 38187584

  View details for PubMedCentralID PMC10769263

• Domain adaptive neural networks improve cross-species prediction of transcription factor binding. Genome research Cochran, K., Srivastava, D., Shrikumar, A., Balsubramani, A., Hardison, R. C., Kundaje, A., Mahony, S. 1800

  Abstract

  The intrinsic DNA sequence preferences and cell-type specific cooperative partners of transcription factors (TFs) are typically highly conserved. Hence, despite the rapid evolutionary turnover of individual TF binding sites, predictive sequence models of cell-type specific genomic occupancy of a TF in one species should generalize to closely matched cell types in a related species. To assess the viability of cross-species TF binding prediction, we train neural networks to discriminate ChIP-seq peak locations from genomic background and evaluate their performance within and across species. Cross-species predictive performance is consistently worse than within-species performance, which we show is caused in part by species-specific repeats. To account for this domain shift, we use an augmented network architecture to automatically discourage learning of training species-specific sequence features. This domain adaptation approach corrects for prediction errors on species-specific repeats and improves overall cross-species model performance. Our results demonstrate that cross-species TF binding prediction is feasible when models account for domain shifts driven by species-specific repeats.

  View details for DOI 10.1101/gr.275394.121

  View details for PubMedID 35042722