FineDNA: Open data to train LLMs on DNA

Datasets:

HuggingFaceBio /

FineDNA

Like 6

Follow

Hugging Face Biology Research 335

Tasks:

Text Generation

Modalities: Text

Formats: parquet

Languages: English

Size: 100M - 1B

Tags: biology

genomics

dna

midtraining

sft

rl

Libraries: Datasets

Dask

Polars

• 1

License: apache-2.0

Dataset card Data Studio Files Files and versions xet

Community 1

This 1115 bp genomic region from Arabidopsis thaliana encompasses the long non-coding RNA (lncRNA) gene AT4G07665 and its surrounding regulatory elements. The locus features a putative TATA box promoter upstream, a centrally located CpG island embedded within the lncRNA body, and multiple microRNA target sites for miR-200c-3p and miR-10b-5p. The presence of several splice acceptor sites both upstream and downstream suggests complex alternative splicing of this non-coding transcript.\n\n<|fim_suffix|>\n\nBeginning my analysis of this 1115-bp sequence from the Arabidopsis genome, I will systematically evaluate the features from the 5' to 3' end. Given that the overall context is annotated as an intergenic region spanning the entirety of the sequence, I expect to find regulatory elements and non-coding features rather than canonical protein-coding structures.\n\nScanning the upstream 5' region, the sequence is generally AT-rich, characteristic of plant promoter and intergenic spaces. At position 345 to 351, I identify a distinct TATAAAA motif. This canonical TATA box serves as a strong core promoter element, indicating the likely presence of a transcription start site slightly downstream. This helps establish the regulatory framework for the region. Shortly after the TATA box, at positions 364 to 365, I observe an AG dinucleotide that is characteristic of a splice acceptor site, complete with a preceding polypyrimidine tract. Its presence upstream of the main gene body suggests complex transcriptional dynamics, potentially indicating that the primary transcript here might be alternatively spliced or that there is an upstream exon not fully captured by the main gene bounds.\n\nMoving further downstream, I encounter the core gene body of a long non-coding RNA (lncRNA) locus, AT4G07665, which is situated on the positive strand and spans positions 436 to 738. Exploring the internal sequence of this lncRNA reveals several critical regulatory features. At positions 532 to 538, the sequence contains the motif CAGTATT. This 7mer-m8 seed match corresponds to a target site for miR-200c-3p. The presence of a microRNA target site within a lncRNA strongly implies that this transcript may function as a competitive endogenous RNA (ceRNA) or miRNA sponge, sequestering miR-200c-3p and thus modulating the expression of its other downstream targets.\n\nDirectly following this miRNA target, from positions 541 to 660, I note a distinct shift in nucleotide composition, forming a CpG island. Although the GC content is moderately elevated at 45% (with an observed-to-expected CpG ratio of 0.86), this localized concentration of CpG dinucleotides within an otherwise AT-biased plant genome is highly significant. CpG islands in lncRNA bodies or promoters are often subject to DNA methylation and act as epigenetic control centers. Embedded within the 5' end of this CpG island, at positions 560 to 565, lies an ACGGGT motif. This corresponds to an 8mer target site for another microRNA, miR-10b-5p. The physical overlap between a CpG island and a microRNA response element raises intriguing possibilities about the interplay between transcriptional or epigenetic silencing and post-transcriptional regulation. Towards the 3' end of the lncRNA body, at positions 699 to 704, the sequence AAGCTT appears, forming a canonical HindIII restriction endonuclease cleavage site.\n\nExiting the 3' boundary of the AT4G07665 lncRNA at position 738, the downstream sequence continues to reveal elements suggestive of complex RNA processing. I detect a series of AG splice acceptor sites scattered through the 3' region. The first of these occurs at positions 781 to 782. Continuing towards the 3' end, two additional splice acceptors are located at positions 911 to 912 and 945 to 946. All of these possess the necessary upstream polypyrimidine sequence context.\n\nSynthesizing these observations into a coherent biological picture, this genomic region represents a highly regulated non-coding transcriptional locus centered around the lncRNA AT4G07665. The presence of a well-defined TATA box promoter drives the transcription of this lncRNA. Its functional significance is deeply tied to its sequence features: rather than coding for a protein, AT4G07665 likely operates as a regulatory RNA. The dual microRNA target sites for miR-200c-3p and miR-10b-5p indicate a role in post-transcriptional networks, potentially acting as a decoy or sponge that dynamically regulates the availability of these miRNAs in specific plant tissues or under stress conditions. Furthermore, the embedded CpG island suggests that the locus itself is under epigenetic control, possibly subject to dynamic methylation that alters its transcription rate in response to developmental cues. Finally, the abundance of splice acceptor sites both upstream and downstream of the primary lncRNA body points to extensive alternative splicing. This locus might produce multiple distinct lncRNA isoforms, each retaining or excluding specific regulatory motifs and miRNA binding sites, thereby fine-tuning its function within the complex regulatory landscape of Arabidopsis.\n\nQ1: How does the location of the CpG island relate to post-transcriptional regulatory elements in this sequence, and what might this imply for the lncRNA's function?\nA1: The CpG island spans positions 541 to 660, directly following the miR-200c-3p target site at 532-538. Furthermore, it completely encompasses the miR-10b-5p target site located at 560-565. This physical overlap suggests that epigenetic modifications, such as DNA methylation at the CpG island, could regulate the expression of the lncRNA and consequently alter the availability of its microRNA binding sites. Thus, the transcript's role as a potential microRNA sponge could be tightly controlled by epigenetic mechanisms.\nQ2: What sequence features suggest that this genomic region may undergo complex RNA processing or alternative splicing?\nA2: The sequence contains multiple AG dinucleotides annotated as splice acceptors with polypyrimidine tracts. One of these is located upstream of the primary lncRNA gene body at positions 364-365. Additionally, there are three more splice acceptors in the downstream sequence at positions 781-782, 911-912, and 945-946. The presence of these multiple splice sites scattered throughout the locus suggests that the region produces alternatively spliced transcripts, which may dynamically include or exclude different regulatory elements.\nQ3: Based on the upstream promoter elements, what regulates the transcription of the lncRNA AT4G07665, and how is it positioned relative to the transcript?\nA3: The transcription of the lncRNA is likely driven by a core promoter featuring a canonical TATA box (TATAAAA) located at positions 345 to 351. This TATA box is positioned approximately 85 base pairs upstream of the lncRNA gene body's start site at position 436. The close proximity of this strong basal promoter element suggests RNA polymerase II-mediated transcription initiation shortly downstream of the TATA box, establishing the primary transcript.<|fim_middle|>TAAAAAGCCATGAAAAGTCTTTAAAATATGCAATTAAAGTACATGTAAAAAAAAAGACATGGATGGATATATCAGTGAGGCATTTTCATCAAATACCTTATCTTCACTCTGCGAAACAGAGCAACATACCAAAAAGTGTAGGTTTTGGTGTGAACCCAGCAGATGGAGAAGATTTTGGTAAAAAAATAGTGAAGATAGAAGAGATGAAGAAGAAAGGAAGGTGTAGTGATATTGTCAGAGAAGGTGAAGTTGGCCGTTTGATAGTGACGTCCACAAATGGATTTTTGCTTAGGATTGAAGAATGCAGAAGAAGCAGAGAAGTAGGAGAACATAAACCGTTGGCCTATAAAACCTAACAACCTTAGTTTTTTGTCTTCGAAGCAAAACTCCGACAGCAACCGCTGCTGCTTTATCTATTCTCGTTTTTTTTTTGGTTAGACAACGCCATCTCTCCTCCCTCTCTCCCTCTCTCCCTCTACGCAAAGCCAAATATTTATAAGAAAATCATTAAACATAAAAAAATTCAAAATCCAGTATTGGGCCGGGTTGGGCCGCGTAAACGGGTATCCAAATTAGAGGGTCGTTGAACTGAGTCTGACTCACTGAGTTTTGTGATCGAGTCAAACAAAGGTGAGCTTGGGAAATAATGGCGTAGCTCTACTAGCTCAAGTTTGAGCTCTCCCATAATGAGTACATCAAGCTTGTCCTCCATCTCCTGCCTCCTCTGGAGATCTCTCTCATGATGACTTCGAAGGTTTGAAAAACCTAACTATGAATCTTAGGATCTGATTACGTTTGACATGTACGTTAAATTGCACAATTTAGGAGTTTTCCTGAGTTCAATAGAAGGTGACAAATTTGGTTCTTGATTGCTTGAAATTGACAGGAACGGAAGTTGTTCACAGATCTTAGTGTTATTCAATTAGAAAAAGATGGAAACTTTTAGCCTTGTTTTGCTTTGTCCTGAAATTATACTCTCTTTGATTGAAATTTGTGTTAGAGCTATCATCTTTGGGTGTTCTTAATTTGCAAACCATTTTTTACTTTATCACTTAAACTCTCACAAAGTCTTTACGCATTATTATTATTATGGCCACCAAACGACTTTGCAGTAA"},"id":{"kind":"null"},"source":{"kind":"string","value":"gemini_fim"},"metadata":{"kind":"null"}}},{"rowIdx":1,"cells":{"text":{"kind":"string","value":"<|fim_prefix|>This genomic region encompasses the Caenorhabditis elegans C33D3.5 protein-coding gene along with its extensive 5' and 3' flanking sequences. The locus is structurally complex, characterized by a dense cluster of upstream alternative splice sites, multiple internal promoters (TATA boxes), and a prominent intronic CpG island. The gene body is further interrupted by a deeply divergent Tc1A DNA transposon remnant. Post-transcriptional regulation is heavily implicated by the presence of multiple microRNA target sites (including miR-17-5p, miR-21-5p, and an 8mer miR-200c-3p site) positioned both within the genic space and in the 3' downstream sequence, which itself is insulated by ancient hAT-family transposon footprints and low-complexity poly-A tracts.\n\n<|fim_suffix|>\n\nAs I initiate my comprehensive sequence walk along this 10,708 base pair segment of the Caenorhabditis elegans X chromosome, I am immediately struck by the dense array of structural and regulatory motifs in the first kilobase. Scanning the 5' region, my attention is arrested almost instantly at position 44. Here, I observe a distinct AG dinucleotide spanning bases 44 to 45. However, this is not merely a random pair of bases; it is preceded by a pyrimidine-rich sequence tract consisting heavily of cytosines and thymines. This is a classic hallmark of a splice acceptor site. In the context of the C. elegans genome, such an upstream splice acceptor often signifies the presence of an independent transcript, an alternatively spliced 5' untranslated region, or the site of trans-splicing. Nematodes frequently utilize Spliced Leader (SL1 or SL2) trans-splicing to process polycistronic operons or to cap the 5' ends of mRNAs, and the robust polypyrimidine tract facilitates the recruitment of the U2AF heterodimer, an essential step in early spliceosome assembly. \n\nAdvancing slightly further down the nucleotide chain, I arrive at positions 259 to 260, where the sequence reveals a canonical GT dinucleotide. This motif acts as a splice donor consensus site. The juxtaposition of a splice acceptor at 44 and a splice donor at 259 creates a fascinating structural dynamic. It defines a small, localized genomic block—roughly 215 base pairs in length—residing far upstream of our primary target gene's main body. The recognition of this GT site by the U1 small nuclear ribonucleoprotein (snRNP) would be the initiating event in excising the subsequent sequence. Proceeding to position 460, I encounter another AG splice acceptor site at 460-461, once again accompanied by an upstream polypyrimidine tract. In C. elegans, intronic lengths are characteristically short, often clustering around 50 base pairs. The distance from the preceding GT donor at 259 to this acceptor at 460 is approximately 200 nucleotides, which is well within the acceptable limits for nematode introns but slightly longer than the median, suggesting a complete intronic removal event bridging the upstream elements.\n\nThe plot thickens as I traverse another short stretch of DNA. At positions 632 to 633, yet another AG splice acceptor emerges. Remarkably, a mere 38 b