# Ensembl Expression Atlas Troubleshooting When Expression Atlas returns no data, the cause is usually not missing biology. In most cases the query itself is the problem. Either the gene was submitted in a format the Atlas does not index, or a filter excludes the only experiments that measured it. The first thing to check is the gene identifier. Expression Atlas indexes genes by stable Ensembl gene IDs and recognized symbols, so a RefSeq accession, a retired Ensembl ID copied from an older paper, or a clone-based name will often return an empty result set even for genes with extensive expression evidence. Expression Atlas is the open expression knowledgebase built and hosted by EMBL-EBI at https://www.ebi.ac.uk/gxa/home. It aggregates gene and protein expression across tissues, cell types, conditions, and species, and the current release comprises more than 4,500 studies from 67 species, including hundreds of single-cell RNA-seq experiments spanning 21 species and an expanding proteomics layer [1]. That scale brings structure, and structure creates specific, predictable ways for a query to fail. This guide walks through those failure modes in order of how often they occur, explains the mechanism behind each one, gives a diagnostic sequence you can complete in minutes, and shows how to cross-check any empty result in the [Ensembl genome browser](/blog/guides/ensembl-genome-browser) itself. If you are new to the resource, the cross-checking section doubles as a short Ensembl [genome browser tutorial](/blog/guides/genome-browsers-for-researchers-a-guide-to-inspecting-genomic-evidence), and the same identifier habits fix problems in both services. ## What Expression Atlas Contains (and What It Does Not) Expression Atlas answers one question: where, at what level, and under what conditions a gene or protein is expressed. Content arrives through curated public studies that the EMBL-EBI team reanalyzes under standardized workflows, with pipeline updates aimed at reproducibility [1]. This matters for troubleshooting, because a gene can only appear if a curated study measured it. The data split into two experiment classes. Baseline experiments report expression under normal or reference conditions. The human baseline draws heavily on Genotype-Tissue Expression (GTEx) tissue profiles, which the current release updated [1]. Differential experiments report genes that change significantly between contrasted conditions, such as treated versus untreated samples. Three content layers sit on top of that split. Bulk RNA-seq accounts for most studies. Single-cell RNA-seq experiments now span 21 species and include externally analyzed community datasets such as Tabula Sapiens and GTEx single-nucleus profiles, which the Atlas curates while preserving their original analytical framework [1]. A marker gene analysis module for bulk baseline experiments identifies genes enriched in specific tissues, which is useful when your gene of interest is absent but a tissue-relevant alternative is needed [1]. Proteomics adds mass spectrometry evidence, with coverage that has increased release over release [1]. Much of that protein evidence originates in PRIDE, the largest mass spectrometry proteomics repository and a founding member of the ProteomeXchange consortium, which disseminates its data to Expression Atlas, Ensembl, and UniProt [2]. Integration matters for troubleshooting. Expression Atlas data flow into Ensembl, UniProt, and Europe PMC, and the resource collaborates with model organism communities such as FlyBase and Gramene [1]. Translational use cases can also draw on the Open Targets platform, which integrates Atlas data [1]. In practice this means an Ensembl gene page can link you to Atlas expression, and an Atlas page can link you back to genome context. What the Atlas does not contain is just as important for diagnosing an empty result. It covers a defined species list, not every published organism. It covers curated studies, not every dataset in public repositories. And it applies expression cutoffs, so genes below the threshold stay invisible even when a study measured them. ## Why Expression Atlas Shows No Data: The Five-Minute Diagnostic ![Eight-step diagnostic sequence from confirming species to going to source data, with the fix for each failure point.](/article-images/illustrations/ensembl-expression-atlas-no-data-five-minute-diagnostic.webp) Run this sequence before anything else. Most no data cases resolve at step 2 or 3. 1. Confirm the species. Open the Expression Atlas home page and check that your organism appears in the species facet. No identifier fix helps when the organism is absent. 2. Resolve the gene to a current Ensembl gene ID. Type the gene symbol into the search bar at www.ensembl.org, select the species match, and copy the stable ID, which begins with a species prefix such as ENSG for human genes. 3. Search the stable ID in Expression Atlas. Identifiers, not symbols, are the reliable key. 4. Open both views. Check the baseline tab and the differential tab separately, because a gene can appear in one and not the other. 5. Adjust the expression cutoff. Baseline results apply a minimum expression threshold, and the results page provides a control to change it. Lower it before concluding anything. 6. Remove filters one at a time. Clear tissue, cell type, and experiment type selections, then reapply them incrementally to find which filter empties the result. 7. Cross-check in the Ensembl genome browser. If Ensembl shows the gene model but the Atlas shows nothing, the gene lacks expression evidence in the included studies rather than failing to exist. 8. Go to the source data. For protein evidence, query PRIDE directly [2]. For transcript evidence in species the Atlas does not cover, use the organism-specific resources described under Cause 2. Each step feeds the next. If step 2 fails because the symbol is unrecognized, the fix is under Cause 1. If step 1 fails because the species is missing, the fix is under Cause 2. ## Quick-Reference Troubleshooting Table | Symptom | Likely cause | How to confirm | Fix | |--|--|--|--| | Symbol search returns nothing | Symbol unrecognized, ambiguous, or not in the Atlas synonym set | Search the symbol at ensembl.org and note the stable ID | Search Expression Atlas with the Ensembl gene ID | | RefSeq accession returns nothing | The Atlas keys on Ensembl identifiers, not RefSeq transcripts | Convert the accession through Ensembl cross-references | Use the converted Ensembl gene ID | | Ensembl ID from an older paper returns nothing | ID retired or merged in a later release | Run the ID through the Ensembl ID History Converter | Search with the successor ID | | Species missing from the search facet | Organism outside the covered species set [1] | Check the species list on the Atlas home page | Use Ensembl Plants, Gramene, or WormBase ParaSite for helminths [3] | | Gene exists in Ensembl, Atlas shows no baseline | No included study detected expression above the cutoff | Open the Ensembl gene page and follow its expression links | Lower the cutoff, or pull source data from primary repositories | | Differential view empty, baseline populated | Gene not significantly changed in the included contrasts | Compare the contrast lists on the experiment pages | Use the baseline view, or download full differential data | | Baseline view empty, gene known to be expressed | Default expression cutoff hides low signal | Adjust the cutoff control on the results page | Lower the cutoff and recheck | | Proteomics filter empties the list | Protein evidence is sparser than transcript evidence [1] | Remove the filter and compare result counts | Query PRIDE directly for protein evidence [2] | | Single-cell view shows nothing | Gene undetected in the selected cell type or dataset | Switch to bulk baseline and compare | Check cell-type metadata before concluding absence | | API returns empty or an error | Endpoint or parameter drift as programmatic access is modernized [1] | Log status codes and read release notes | Pin endpoints, or use bulk download files | | Coordinates from a paper do not match the browser | Assembly build mismatch, such as GRCh37 era coordinates [4] | Check the assembly label on the gene page | Search by gene ID, or lift coordinates to the current build | ## Cause 1: Gene Identifier Problems Identifier mismatch is the most common reason Expression Atlas returns no data, and it produces the most misleading symptom, because the gene may be richly documented everywhere else. ### Symbols are labels, not keys Gene symbols are convenient for humans and unreliable for machines. They can be ambiguous across species, they can change as nomenclature evolves, and some gene classes carry names that look nothing like genes. Long intergenic non-coding RNAs frequently carry clone-based names derived from genomic clone plates rather than from any biological property, which makes symbol search fragile. Expression Atlas resolves recognized symbols, but a symbol that is outdated, ambiguous, or absent from its synonym list returns an empty result. Stable Ensembl gene IDs behave differently. The identifier persists across releases, changes are tracked, and every resource that integrates with Ensembl can resolve it [1]. Treat symbols as display labels and stable IDs as database keys, in the web interface and in scripts alike. ### Retired and merged stable IDs Stable does not mean immortal. Ensembl releases increment through the year, and across a decade of updates gene models merge, split, and retire. The span is easy to see in the literature. One heart failure transcriptome study cross-referenced its noncoding RNAs to Ensembl release 73 on the GRCh37 assembly [4]. An ID copied from a release 73 era paper may map to a different gene today, or to nothing at all. The Ensembl website provides an ID History Converter for exactly this case. Enter the old ID, and the tool lists successor IDs together with the release in which each change occurred. Merged IDs map forward to the surviving gene. Retired IDs without a successor indicate a gene model that no longer exists in current annotation, which changes the troubleshooting path entirely, because no identifier will return data for a gene the annotation no longer contains. At that point the search belongs under Cause 2. ### Converting identifiers programmatically For single queries, the Ensembl search box is enough. For pipelines, use the REST API. The cross-reference endpoint resolves a symbol to identifiers. ``` curl "https://rest.ensembl.org/xrefs/symbol/homo_sapiens/BRCA1?content-type=application/json" ``` The lookup endpoint expands an ID into the full gene object, including location, biotype, and current symbol. ``` curl "https://rest.ensembl.org/lookup/id/ENSG00000230615?content-type=application/json" ``` BioMart, linked from the Ensembl website, handles bulk conversion of hundreds to thousands of identifiers and returns a table you can join against your own data. Space requests to respect the service's fair-use limits, and log responses so a silent empty result is distinguishable from a network failure. A short safety note for this section. Review any third-party conversion script before running it, verify downloaded files against checksums where publishers provide them, and keep human donor-derived expression files, such as GTEx downloads, on access-controlled storage in line with the access terms attached to each study. ## Cause 2: Species and Annotation Coverage ### Your species may not be included Expression Atlas covers 67 species [1]. Check the species facet before anything else, because coverage is deepest for human, mouse, and major model and crop organisms, and no query trick extends it. The resource collaborates with model organism communities such as FlyBase and Gramene to broaden reach [1], but the covered list remains finite. Absence from the Atlas does not mean absence from the Ensembl ecosystem, and matching your organism to the right Ensembl-dependent resource is often the actual fix for a no data result. Parasite researchers have a dedicated path. WormBase ParaSite provides genomes, consistent functional annotation, comparative analysis, and [gene expression](/blog/guides/gene-expression) analysis for more than 100 helminth species, is developed in collaboration with WormBase and Ensembl Genomes, and releases three to six times per year [3]. Plant researchers commonly start in Ensembl Plants, where gene family workflows retrieve stable IDs and sequences for downstream expression analysis [5]. ### The gene may be absent from Ensembl annotation The second coverage failure is subtler. Your species may be covered, but your gene may not exist in the Ensembl gene set that the Atlas is keyed to. Gene content varies across databases because of differences in [RNA sequencing](/blog/guides/rna-sequencing) data and analysis pipelines, and the variation is largest for long non-coding RNAs, which have higher tissue and developmental specificity and are harder to identify consistently than protein-coding genes [6]. The chicken GRCg7b assembly shows the scale of the gap. The Ensembl annotation listed 17,007 protein-coding genes, while an integrated atlas built from RefSeq, Ensembl, and other resources reached 24,102. For lncRNAs the difference was wider, 11,944 in Ensembl against 44,428 in the integrated set [6]. Expression evidence also falls short of annotation. Across 47 tissues, expression supported 93 percent of protein-coding genes but only 79 percent of lncRNAs in that integrated set [6]. Expression studies inherit these limits, because analyses estimate expression against a chosen annotation release and then apply robust expression filters. A cartilage lncRNA study working from GENCODE and Ensembl annotations identified 5,053 robustly expressed lncRNAs, of which 191 were significantly differentially expressed between lesioned and preserved tissue [7]. A gene can therefore be real, annotated, and absent from every expression result you can query, simply because no included study detected it above background in the tissue you care about. The BLUEPRINT hematopoietic atlas shows the same dependency from the quantification side, estimating gene and isoform levels against existing Ensembl annotations [8]. ### Annotation drift between releases Even when the gene exists, its model changes across releases. Ensembl version 109, from 2023, carried approximately 252,798 distinct RNA isoform annotations across 62,710 gene bodies [9]. Isoform catalogs grow with each release, and quantification against a newer annotation can detect transcripts an older study never measured. When comparing your own quantification against Atlas results, match annotation releases where possible, and at minimum record which release each dataset used. ## Cause 3: Experiment Type, Filters, and Cutoffs ### Baseline versus differential An empty differential view with a populated baseline view is not a failure. The two views answer different questions. Baseline reports where a gene is expressed under reference conditions, drawing on resources such as the updated GTEx tissue profiles in human [1]. Differential reports genes that change significantly between contrasted conditions. A gene expressed in every tissue may never pass a differential significance test, and a condition-specific gene may show near-zero baseline in most tissues. Check both tabs before concluding a gene has no data, and read the experiment descriptions to see which contrasts were actually tested. ### The expression cutoff Baseline results apply a minimum expression threshold, and genes below it are hidden. The results page includes a cutoff control. Lower the cutoff and recheck before concluding absence. This single action resolves a large share of no data reports for lowly expressed genes, including many transcription factors and non-coding RNAs whose biological signal sits close to the detection floor. ### Proteomics filters Protein-level evidence is sparser than transcript-level evidence. Expression Atlas has increased proteomics coverage release over release [1], but mass spectrometry detection remains narrower than RNA-seq measurement, so filtering results to proteomics removes most genes. When a protein view comes back empty, remove the filter to see the transcriptomic picture, then query PRIDE directly for protein evidence, since PRIDE disseminates its processed data to Expression Atlas, Ensembl, and UniProt [2]. ### Single-cell resolution Single-cell experiments resolve expression by cell type, which changes what no data means. Hundreds of single-cell RNA-seq experiments across 21 species are included, among them externally analyzed community datasets such as Tabula Sapiens and GTEx single-nucleus profiles [1]. A gene absent from a selected cell type may be strongly expressed in another. Switch between bulk and single-cell views, inspect the cell-type metadata, and scan the full cell-type list before concluding that a gene was never detected. ## Cause 4: Query and API Failures ### Web interface failures Browser-side problems are rare but real. Cached pages can serve stale results after a release update, aggressive content blockers can break the interactive heatmaps, and very large queries can time out. Hard refresh the page, disable blockers for the domain, and narrow the query scope, for example by filtering to one tissue before expanding outward. ### REST API failures Programmatic access fails differently from the web interface. The current release report notes that enhancing programmatic access is an active direction [1], which means endpoints and response formats can change between releases. Three habits prevent most API failures. Log the HTTP status code with every response, so an empty body is distinguishable from a routing error. Pin the endpoint and parameter set in your scripts, and reread release notes after each Ensembl and Expression Atlas update. Test with a gene you know has data, so a failure is attributable to the query rather than to the gene. The two endpoints shown under Cause 1, xrefs for symbol resolution and lookup for ID expansion, are the workhorses for identifier hygiene. An empty list from a symbol search is diagnostic in itself. It means Ensembl does not recognize the symbol in that species, which points back to Cause 1. ### Bulk downloads as the fallback When interactive or programmatic access fails, the Download section of the Expression Atlas site provides the underlying expression matrices. Bulk files suit reproducible pipelines better than scraped web results in any case. Download the species and experiment files you need, record the download date and release, and verify file integrity before analysis. A short safety note applies here as well. Do not pipe downloaded scripts or data directly into execution without inspection, and store donor-derived human datasets according to their access terms. A pipeline that depends on a live API will break eventually. A pipeline built on versioned snapshots will not. ## Cause 5: Cross-Species and Homology Lookups A frequent variant of the no data problem is cross-species. You have expression data for a gene in one species and want the counterpart in another. The mapping between species runs through Ensembl homology tables, which classify relationships as one-to-one, one-to-many, or many-to-many [10]. The classification matters because default many-to-many mappings can overweight expanded gene families. In cross-species single-cell integration, many-to-many mappings generated mapping-associated microclusters that lacked clear cell-type identity, while restricting to one-to-one mappings suppressed the artifacts at the cost of reducing homologous pairs by roughly 8 percent, about 900 pairs [10]. The practical rule for troubleshooting is direct. If your gene appears to have no ortholog with expression data, first check whether the relationship is one-to-many. The counterpart may exist as a family rather than a single gene, and expression may attach to only one family member. Prefer one-to-one orthologues for cross-species expression comparison, and treat one-to-many relationships as a set to be examined member by member. ## Cross-Checking in the Ensembl Genome Browser: A Short Tutorial When Expression Atlas returns nothing, the Ensembl genome browser tells you whether the problem is the gene or the expression evidence. The check takes minutes, and if you are learning how to use Ensembl for the first time, these steps build the habits that keep expression queries working. 1. Open www.ensembl.org and enter the gene symbol or ID in the search bar. The search returns matches across species and identifier types. 2. Select the species match and open the gene page. The page header shows the stable ID, symbol, biotype, and the assembly with coordinates. 3. Read the assembly label. Human pages currently sit on GRCh38. Coordinates and IDs from GRCh37 era papers will not line up with current pages, so search by ID rather than by coordinate when crossing builds. 4. Scan the transcript table. Current annotation is dense, and transcript-level views can bury a simple question. Confirm your gene is present, note its biotype, and work at the gene level for expression checks, because non-coding biotypes carry thinner expression evidence. 5. Follow the expression links. Expression Atlas data are integrated into Ensembl [1], and the gene page connects to expression resources. If the link resolves, your identifier is correct and the gap is in expression evidence. If the gene page itself is missing, the problem is annotation, not expression. 6. Open the orthologues view for cross-species questions. The page lists one-to-one, one-to-many, and many-to-many relationships [10], which tells you whether a missing counterpart is truly absent or merely split across a family. 7. Use BioMart for bulk work. BioMart exports IDs, symbols, biotypes, and homology tables for hundreds to thousands of genes at a time, which suits pipeline-level identifier hygiene better than page-by-page checks. The outcome of this check sorts every no data case into two buckets. A gene present in Ensembl with expression links means the Atlas query needs adjustment. A gene absent or retired means the annotation itself is the limit, and the search should move to organism-specific resources or source repositories. ## Worked Examples ### Example 1: Resolving a lincRNA by stable ID A breast cancer lincRNA study illustrates the identifier path end to end. The study profiled 1,000 tumor transcriptomes from The Cancer Genome Atlas in its discovery stage, identified 83 lincRNAs with a false discovery rate below 1 percent, and validated 37 of them in a replication cohort [11]. One of the validated lincRNAs, reported under the clone-based name RP5-1198O20 with the Ensembl ID ENSG00000230615, showed an association with breast cancer survival [11]. Searching RP5-1198O20 in an expression resource is unreliable, because clone-based names resolve poorly across databases. Searching ENSG00000230615 resolves determinately. The lookup command shown earlier returns the current gene object for that ID, confirming the gene exists in current annotation. Whether Expression Atlas carries expression for it then depends on the included studies and filters, which is exactly the boundary between Cause 1 and Cause 3. ### Example 2: A plant gene family through Ensembl Plants Plant workflows show the species-coverage fix in action. A study of the Swi2/Snf2 [chromatin remodeling](/knowledge/molecular-biology/chromatin-remodeling) family in maize identified 35 family members, named ZmCHR1 through ZmCHR35, using the Ensembl Plants database, then characterized domain architecture, chromosomal distribution, and expression [5]. Expression analysis indicated tissue-specific patterns and differential expression under stress conditions, and quantitative PCR under drought showed certain genes upregulated at 12 hours and downregulated at 24 hours [5]. The workflow generalizes to any crop or non-model plant. Retrieve stable IDs and sequences from Ensembl Plants, characterize the gene set, then query expression resources where the species is covered, falling back to the study's own experimental validation where it is not. When the species is absent from Expression Atlas, the Ensembl Plants entry point still anchors every downstream identifier. ### Example 3: A developmental baseline in zebrafish Baseline coverage questions are easiest to see in a model organism. A zebrafish study produced an mRNA expression time course across 18 developmental time points from the one-cell stage to 5 days post [fertilization](/blog/guides/fertilization), profiling 23,642 genes, and the dataset is browsable and downloadable at both Expression Atlas and Ensembl [12]. A zebrafish gene that returns nothing under a default query may simply sit outside the sampled stages or below the cutoff for those stages. Narrow the stage selection, lower the cutoff, and confirm the gene falls within the profiled set before concluding the resource lacks the data. ## Common Mistakes and Limitations ![Decision path asking whether the species is covered, the ID resolves, and the gene is in Ensembl, ending at the right fix.](/article-images/illustrations/ensembl-expression-atlas-no-data-empty-result-triage.webp) Several habits turn a simple empty result into a wrong conclusion. - Reading absence as absence. An empty result means no included study detected the gene above the applied cutoff in the selected context. It does not mean the gene is silent. - Confusing the two services. Ensembl is the genome database and browser. Expression Atlas is the expression knowledgebase built on curated studies. They integrate [1], but a gene can be fully present in one and absent from the other. - Ignoring release versions. Annotation changed substantially between the GRCh37 era and current releases. IDs, transcript counts, and results are all release specific, and mixing releases silently corrupts comparisons. - Treating many-to-many homology as orthology. Expanded gene families need member-by-member examination, and one-to-one restriction is the safer default for cross-species expression claims [10]. - Querying with old coordinates. GRCh37 era coordinates do not match current assembly pages, so coordinate-based searches across builds fail quietly. - Over-reading baseline expression. Baseline level and location describe where a gene is expressed, not how it is regulated or what it does. - Generalizing from one resource. Downstream databases harvest Ensembl annotations and expression evidence with their own filters, so a gene missing from one resource can be well supported in another. One limitation deserves plain statement. Individual datasets, particularly those touching disease association or clinical interpretation, need review by a specialist familiar with the organism, the assay, and the statistics before any conclusion is drawn from an empty or populated result. ## Prevention Practices Identifier hygiene prevents most recurrence. - Record the Ensembl release and genome assembly in every methods section and script header. - Key pipelines on stable gene IDs, and carry symbols only as display labels. - Run ID history checks after each Ensembl release if your gene set is large or includes older literature IDs. - Maintain a local mapping table of IDs, symbols, biotypes, and release versions, and refresh it on a schedule. - Download versioned snapshots of expression data for reproducible analysis instead of depending on live queries. - Log API endpoints, parameters, and response codes, so a future failure has a paper trail. - When starting a new organism or gene family, confirm species and annotation coverage before designing the analysis, the way plant gene family studies confirm the Ensembl Plants entry point first [5]. These practices cost minutes and prevent the most expensive failure mode in expression work, which is building an analysis on identifiers that quietly changed underneath it. ## Frequently Asked Questions ### Why does Expression Atlas show no data for my gene? The most common reason is an identifier mismatch. Expression Atlas indexes genes by stable Ensembl gene IDs and recognized symbols, so RefSeq accessions, retired IDs, and clone-based names return empty results. Resolve your gene to its current Ensembl ID and search again. ### How do I convert an old Ensembl gene ID to the current one? Use the ID History Converter on the Ensembl website. IDs that were retired or merged across releases map forward to successor IDs, and the tool lists each change with the release in which it occurred. ### Does Expression Atlas cover every species? No. Check the species list on the home page first. For parasites, WormBase ParaSite provides expression analysis for helminth genomes, and plant researchers can work through Ensembl Plants and Gramene. ### Why does my gene appear in Ensembl but not in Expression Atlas? Ensembl holds the gene model, while Expression Atlas holds expression evidence from curated studies. A gene can be annotated yet lack expression evidence in the tissues and conditions the Atlas includes, which is common for long non-coding RNAs. ### What is the difference between baseline and differential expression in Expression Atlas? Baseline experiments report where and at what level a gene is expressed under reference conditions, such as across human tissues. Differential experiments report genes that change significantly between contrasted conditions. A gene can appear in one view and not the other. ### Why do proteomics filters remove most of my results? Protein expression evidence is smaller in scale than [transcriptomics](/knowledge/bioinformatics/modern-transcriptomics-bulk-single-cell-spatial) evidence. Mass spectrometry detections remain sparser than RNA-seq measurements, so filtering to proteomics narrows the gene list sharply. Remove the filter or query PRIDE directly. ### Should I use gene symbols or Ensembl IDs in scripts? Use stable Ensembl gene IDs in scripts and store symbols as display labels. Symbols can be ambiguous or change over time, while stable IDs persist across releases, with any retirements tracked in the ID history. ### What should I do if the website works but the API returns nothing? Check the endpoint, parameters, and release notes first, because programmatic access is being modernized and endpoints can change. As a fallback, use the bulk download files on the Expression Atlas site, which mirror the same data. ## Sources 1. [Expression Atlas in 2026: enabling FAIR and open expression data through community collaboration and integration.](https://pubmed.ncbi.nlm.nih.gov/41370097/) 2. [The PRIDE database and related tools and resources in 2019: improving support for quantification data.](https://pubmed.ncbi.nlm.nih.gov/30395289/) 3. [Using WormBase ParaSite: An Integrated Platform for Exploring Helminth Genomic Data.](https://pubmed.ncbi.nlm.nih.gov/29761467/) 4. [Right ventricular long noncoding RNA expression in human heart failure.](https://pubmed.ncbi.nlm.nih.gov/25992278/) 5. [Exploring the Roles of the Swi2/Snf2 Gene Family in Maize Abiotic Stress Responses.](https://pubmed.ncbi.nlm.nih.gov/39273633/) 6. [Enriched atlas of lncRNA and protein-coding genes for the GRCg7b chicken assembly and its functional annotation across 47 tissues.](https://pubmed.ncbi.nlm.nih.gov/38504112/) 7. [Elucidating Epigenetic Regulation by Identifying Functional cis-Acting Long Noncoding RNAs and Their Targets in Osteoarthritic Articular Cartilage.](https://pubmed.ncbi.nlm.nih.gov/32840049/) 8. [Cell type-specific novel long non-coding RNA and circular RNA in the BLUEPRINT hematopoietic transcriptomes atlas.](https://pubmed.ncbi.nlm.nih.gov/32703790/) 9. [A bioinformatic survey of RNA isoform diversity and expression across 9 GTEx tissues using long-read sequencing data.](https://pubmed.ncbi.nlm.nih.gov/41345545/) 10. [Protein large language model-assisted one-to-one gene homology mapping in cross-species single-cell transcriptome integration.](https://pubmed.ncbi.nlm.nih.gov/42855280/) 11. [Long intergenic non-coding RNA expression signature in human breast cancer.](https://pubmed.ncbi.nlm.nih.gov/27897201/) 12. [A high-resolution mRNA expression time course of embryonic development in zebrafish.](https://pubmed.ncbi.nlm.nih.gov/29144233/)