Review Maps Tools for Finding Peptides in Noncoding RNA

A new review published in the journal Amino Acids provides a comprehensive overview of the computational tools, databases, and in silico strategies used to identify peptides encoded by noncoding RNAs. The study, led by Nidhi Shenoy and colleagues, consolidates resources across major noncoding RNA classes, including long noncoding RNAs, circular RNAs, and primary microRNAs. This review addresses a critical gap in the field by offering a unified framework for evaluating prediction tools, moving beyond previous surveys that focused on individual RNA subclasses. The authors emphasize that these computational methods are essential for uncovering hidden functional peptides within the noncoding transcriptome, which was previously dismissed as transcriptional artifacts.

Noncoding RNAs have gained significant recognition in recent years for their roles in gene regulation and their ability to encode functional molecules known as ncRNA-encoded peptides, or ncPEPs. These peptides are a subset of micropeptides derived from small open reading frames contained in transcripts annotated as noncoding RNAs. Despite their small size, typically fewer than 100 amino acids, ncPEPs are emerging as a new functional class of molecules that contribute to multiple cellular activities. The discovery of these peptides has opened new avenues in proteomics and genomics research, revealing biological mechanisms that were previously unexplored and challenging the traditional view of coding versus noncoding transcripts.

The identification of ncPEPs requires a multifaceted approach that utilizes bioinformatics tools, databases, and experimental methodologies. Bioinformatics tools serve as indispensable aids in scanning genomic and transcriptomic datasets to identify potential small open reading frames within noncoding regions. These tools consider various features, including open reading frame length, nucleotide bias, internal ribosome entry sites, and N6-methyladenosine modification sites. Several interactive and continually updated databases provide comprehensive repositories of experimentally validated and predicted ncPEPs, streamlining annotation efforts and guiding functional research. The review highlights how these resources enable systematic annotation and cross-referencing of candidate peptides across multiple species.

The authors categorize the identification process into three distinct investigative tiers, reflecting the progressive convergence of different methodologies. The first tier relies on sequence-level evidence, utilizing high-throughput RNA sequencing and coding potential estimation to distinguish coding from noncoding transcripts. Methods such as CPAT, CNCI, and PORTRAIT rely on intrinsic sequence characteristics and evolutionary features. However, the review notes that these tools often show reduced sensitivity toward small open reading frames due to weak coding signatures and training biases toward canonical protein-coding genes. Homology-dependent approaches and conservation-based frameworks also have limitations, as they may fail to detect recently evolved or species-specific peptides that lack detectable database matches.

To address the limitations of sequence-based prediction, the second tier focuses on translational evidence, utilizing methods such as ribosome profiling, polysome profiling, and translating ribosome affinity purification. Ribosome profiling captures ribosome-protected RNA fragments to provide a snapshot of global translation, with the discovery of three-nucleotide periodicity marking a significant advancement in distinguishing active translation from nonspecific ribosome binding. Tools such as RiboTaper, Rp-Bp, and RibORF leverage this periodicity to enhance the detection of translated small open reading frames. A specialized variant, global translation initiation sequencing, distinguishes between ribosome initiation and elongation using specific translation inhibitors, providing deeper insights into translation events.

The third tier involves protein-level evidence through proteogenomics, which integrates genomics or transcriptomic data with mass spectrometry-based proteomics. This approach provides direct evidence of peptide translation by creating a custom peptide database from predicted small open reading frames and matching them against mass spectrometry spectra. For example, a study by Sun et al. identified six peptides encoded by ab initio-predicted genes that had not been previously validated, underscoring the utility of proteogenomic approaches in refining human gene annotation. However, challenges remain, as standard mass spectrometry protocols are biased toward more abundant larger peptides, and many small open reading frames could be undetected due to their rapid degradation.

The review also details several publicly available databases that compile experimentally validated and computationally predicted ncPEPs. The ncEP database records information on peptides translated by noncoding RNAs, curated manually from research articles, covering peptides across several species with their function and experimental evidence. SPENCER provides a platform for exploring functional ncPEPs in cancer derived from mass spectrometry data, categorizing peptides across 15 different cancer types. FuncPEP focuses on ncPEPs with validated functional roles in cellular processes or disease contexts, while TransLnc is specifically focused on translatable lncRNA peptides in humans, rats, and mice. LncPEP provides comprehensive data on lncRNA-encoded peptides across 39 species, combining five different lines of evidence including CPC2, CPAT, Ribo-seq, m6A, and translation initiation site prediction.

Additional databases such as smProt catalog small proteins derived from untranslated regions and noncoding regions, compiling data from Ribo-seq predictions, literature curation, existing databases, and mass spectrometry evidence across eight species and 291 cell lines. NCPbook includes a vast collection of noncanonical peptides from 29 species, supported by evidence from mass spectrometry, molecular experiments, or ribosome profiling. Complementary circRNA annotation resources, such as circRNADb, provide important upstream support for circRNA-encoded peptide discovery by annotating genomic coordinates, exon splicing information, predicted open reading frames, and internal ribosome entry site elements. These resources are essential for researchers seeking to identify candidate circRNA-encoded peptides prior to experimental validation.

The functional roles of ncPEPs span a wide array of biological processes, including developmental processes, core cellular functions, immune responses, and disease progression. In developmental processes, the first animal-derived ncPEPs were identified in Drosophila, where lncRNA-encoded peptides are necessary for the development of flies during the embryonic stage. In zebrafish, the Toddler peptide functions as a motogen that encourages cell migration by triggering G protein-coupled APJ signaling. In core biological functions, peptides such as Mitoregulin play a fundamental role in cellular energy production and homeostasis, while Humanin, encoded within mitochondrial rRNA, regulates apoptosis by inhibiting Bax translocation to mitochondria.

In immune and disease responses, lncRNA-encoded ncPEPs have emerged as significant diagnostic and prognostic markers in various cancers. HOX-AS3, a conserved 53 amino acid peptide, prevents tumorigenesis by controlling PKM alternative splicing and metabolic reprogramming in colon cancer cells. NoBody, a 68 amino acid peptide, interacts with proteins involved in mRNA decapping and decay and is inversely related to the expression levels of mutant oncogenes. Several circRNA-encoded peptides, such as SHPRH-146aa and PINT-87aa, have been found to be involved in cancers like glioma, hepatocellular carcinoma, and colorectal cancer, often functioning as tumor suppressors by inhibiting cell proliferation or preventing oncogene transcription.

In plant systems, ncPEPs play a role in growth, development, and stress responses. The Enod40 lncRNA, the earliest eukaryotic micropeptide discovered in plant biology, encodes two short peptides involved in the organogenesis of root nodules. The POLARIS gene encodes a 36 amino acid peptide that influences vascular patterning and cell proliferation. Additional ncPEPs, such as ROT4, miPEP858a, and miPEP408, have demonstrated potential in enhancing agricultural traits, including the modulation of flowering time and the promotion of root growth and nodulation. Synthetic peptides such as miPEP171b and miPEP165a have been shown to slow lateral root development and stimulate main root growth, respectively.

The review discusses the evolving strategies for the computational and experimental identification of ncPEPs, highlighting the limitations of current machine-learning approaches. Labeled datasets specific to ncPEPs are still small and noisy, and only a minority of predicted small open reading frames have been biochemically validated. Many lncRNAs appear to exert both RNA and peptide-mediated functions, and there are few curated negative sets, all of which constrain supervised training for coding-potential and small open reading frame detection models. In addition, the Ribo-seq and mass spectrometry data used as training targets contain systematic artifacts such as scanning ribosomes, digestion and coverage biases, and under-detection of small open reading frames, which can propagate into model predictions if not explicitly modeled.

Future directions in ncPEP research are likely to be driven by multi-modal machine learning models that integrate sequence features with translatome and proteogenomic evidence. Tools that combine multiple Ribo-seq-derived scores, such as recent ribosome-profiling-assisted proteogenomics frameworks, already improve sensitivity and confidence for microprotein discovery compared with single-signal callers alone. Deep learning architectures have become state-of-the-art for distinguishing coding from noncoding transcripts, with convolutional and recurrent models consistently outperforming traditional coding-potential indices. The development of comprehensive and well-curated databases remains essential, and the future lies in the use of large models and AI-driven platforms that may accelerate functional annotation and database curation by integrating sequence homology, motif detection, and literature mining.

Despite these advancements, ncPEP research remains in its early stages, with challenges persisting in the detection, validation, and functional characterization of these small peptides. Since ncRNAs can function both as regulatory RNA molecules and translated functional peptides, this additional level of complexity expands the scope of their potential in human disorders. The discovery and functional characterization of ncPEPs represents a rapidly advancing frontier in molecular biology, and translating these discoveries into clinical applications will require the integration of AI-driven annotation frameworks with expert human curation. This ensures accurate interpretation of the hidden proteomic landscape encoded within ncRNAs, highlighting the clinical utility of ncPEPs as tumor biomarkers and therapeutic targets for the early diagnosis and treatment of human illnesses.

Subject of Research: Computational identification and characterization of noncoding RNA-encoded peptides: tools, databases, and in silico strategies

Article Title: Computational identification and characterization of noncoding RNA-encoded peptides: tools, databases, and in silico strategies

Article References: Shenoy, N., Karkare, A., Gangadhar, V., Padubidri, S. R., Rao, S., Dsouza, L. A., Joshi, M. B., Damerla, R. R., Gandhi, N. S., & Mallya, S. (2026). Computational identification and characterization of noncoding RNA-encoded peptides: tools, databases, and in silico strategies. Amino Acids, 58(1), Article 38. https://doi.org/10.1007/s00726-026-03531-3

Image Credits: AI Generated

DOI: 10.1007/s00726-026-03531-3

Keywords: Computational, identification, characterization, noncoding, RNA-encoded, peptides, tools, databases, silico, strategies, scientific research