G-quadruplexes are nucleic acid secondary structures formed by a stack of square planar G-quartets. G-quadruplexes were implicated in many biological functions including telomere maintenance, replication, transcription, and translation, in many species including humans and plants. For wheat, however, though it is one of the world's most important staple food, no G-quadruplex studies have been reported to date. Here, we computationally identify putative G4 structures (G4s) in wheat genome for the first time and compare its distribution across the genome against five other genomes (human, maize, Arabidopsis, rice, and sorghum). We identified close to 1 million G4 motifs with a density of 76 G4s/Mb across the whole genome and 93 G4s/Mb over genic regions. Remarkably, G4s were enriched around three regions, two located on the antisense and one on the sense strand at the following positions: 1) the transcription start site (TSS) (antisense), 2) the first coding domain sequence (CDS) (antisense), and 3) the start codon (sense). Functional enrichment analysis revealed that the gene models containing G4 motifs within these peaks were associated with specific gene ontology (GO) terms, such as developmental process, localization, and cellular component organization or biogenesis. We investigated genes encoding MADS-box transcription factors and showed examples of G4 motifs within critical regulatory regions in the VRN-1 genes in wheat. Furthermore, comparison with other plants showed that monocots share a similar distribution of G4s, but Arabidopsis shows a unique G4 distribution. Our study shows for the first time the prevalence and possible functional roles of G4s in wheat.

Protein phosphorylation is a dynamic and reversible post-translational modification that regulates a variety of essential biological processes. The regulatory role of phosphorylation in cellular signaling pathways, protein-protein interactions, and enzymatic activities has motivated extensive research efforts to understand its functional implications. Experimental protein phosphorylation data in plants remains limited to a few species, necessitating a scalable and accurate prediction method. Here, we present PhosBoost, a machine-learning approach that leverages protein language models and gradient-boosting trees to predict protein phosphorylation from experimentally derived data. Trained on data obtained from a comprehensive plant phosphorylation database, qPTMplants, we compared the performance of PhosBoost to existing protein phosphorylation prediction methods, PhosphoLingo and DeepPhos. For serine and threonine prediction, PhosBoost achieved higher recall than PhosphoLingo and DeepPhos (.78, .56, and .14, respectively) while maintaining a competitive area under the precision-recall curve (.54, .56, and .42, respectively). PhosphoLingo and DeepPhos failed to predict any tyrosine phosphorylation sites, while PhosBoost achieved a recall score of .6. Despite the precision-recall tradeoff, PhosBoost offers improved performance when recall is prioritized while consistently providing more confident probability scores. A sequence-based pairwise alignment step improved prediction results for all classifiers by effectively increasing the number of inferred positive phosphosites. We provide evidence to show that PhosBoost models are transferable across species and scalable for genome-wide protein phosphorylation predictions. PhosBoost is freely and publicly available on GitHub.

Machine learning and modeling approaches have been used to classify protein sequences for a broad set of tasks including predicting protein function, structure, expression, and localization. Some recent studies have successfully predicted whether a given gene is expressed as mRNA or even translated to proteins potentially, but given that not all genes are expressed in every condition and tissue, the challenge remains to predict condition-specific expression. To address this gap, we developed a machine learning approach to predict tissue-specific gene expression across 23 different tissues in maize, solely based on DNA promoter and protein sequences. For class labels, we defined high and low expression levels for mRNA and protein abundance and optimized classifiers by systematically exploring various methods and combinations of k-mer sequences in a two-phase approach. In the first phase, we developed Markov model classifiers for each tissue and built a feature vector based on the predictions. In the second phase, the feature vector was used as an input to a Bayesian network for final classification. Our results show that these methods can achieve high classification accuracy of up to 95% for predicting gene expression for individual tissues. By relying on sequence alone, our method works in settings where costly experimental data are unavailable and reveals useful insights into the functional, evolutionary, and regulatory characteristics of genes.

G-quadruplexes (G4s) are four-stranded nucleic acid structures with closely spaced guanine bases forming square planar G-quartets. Aberrant formation of G4 structures has been associated with genomic instability. However, most plant species are lacking comprehensive studies of G4 motifs. In this study, genome-wide identification of G4 motifs in barley was performed, followed by a comparison of genomic distribution and molecular functions to other monocot species, such as wheat, maize, and rice. Similar to the reports on human and some plants like wheat, G4 motifs peaked around the 5' untranslated region (5' UTR), the first coding domain sequence, and the first intron start sites on antisense strands. Our comparative analyses in human, Arabidopsis, maize, rice, and sorghum demonstrated that the peak points could be erroneously merged into a single peak when large window sizes are used. We also showed that the G4 distributions around genic regions are relatively similar in the species studied, except in the case of Arabidopsis. G4 containing genes in monocots showed conserved molecular functions for transcription initiation and hydrolase activity. Additionally, we provided examples of imperfect G4 motifs.

Background

G-quadruplexes (G4s), formed within guanine-rich nucleic acids, are secondary structures involved in important biological processes. Although every G4 motif has the potential to form a stable G4 structure, not every G4 motif would, and accurate energy-based methods are needed to assess their structural stability. Here, we present a decision tree-based prediction tool, G4Boost, to identify G4 motifs and predict their secondary structure folding probability and thermodynamic stability based on their sequences, nucleotide compositions, and estimated structural topologies.

Results

G4Boost predicted the quadruplex folding state with an accuracy greater then 93% and an F1-score of 0.96, and the folding energy with an RMSE of 4.28 and R² of 0.95 only by the means of sequence intrinsic feature. G4Boost was successfully applied and validated to predict the stability of experimentally-determined G4 structures, including for plants and humans.

Conclusion

G4Boost outperformed the three machine-learning based prediction tools, DeepG4, Quadron, and G4RNA Screener, in terms of both accuracy and F1-score, and can be highly useful for G4 prediction to understand gene regulation across species including plants and humans.

Computational approaches have provided new biological insights into the chemical mechanism of action of cellulases, which are used in the industrial production of bioethanol. Fine-grained methods, such as molecular dynamics and quantum mechanics, as well as coarse-grained methods, such as elastic network models, were used to investigate how the chemistry and structural dynamics of these enzymes contribute to their function. In this review, we highlight recent computational studies to understand this crucial biofuel enzyme class’s chemistry and structural dynamics, as well as their significance in revealing enzymatic mechanism of action. Computational methods can complement and amplify the findings of experimental methods, which can be used in tandem to create more efficient industrial enzymes.

Background

With the advances in the high throughput next generation sequencing technologies, genome-wide association studies (GWAS) have identified a large set of variants associated with complex phenotypic traits at a very fine scale. Despite the progress in GWAS, identification of genotype-phenotype relationship remains challenging in maize due to its nature with dozens of variants controlling the same trait. As the causal variations results in the change in expression, gene expression analyses carry a pivotal role in unraveling the transcriptional regulatory mechanisms behind the phenotypes.

Results

To address these challenges, we incorporated the gene expression and GWAS-driven traits to extend the knowledge of genotype-phenotype relationships and transcriptional regulatory mechanisms behind the phenotypes. We constructed a large collection of gene co-expression networks and identified more than 2 million co-expressing gene pairs in the GWAS-driven pan-network which contains all the gene-pairs in individual genomes of the nested association mapping (NAM) population. We defined four sub-categories for the pan-network: (1) core-network contains the highest represented ~ 1% of the gene-pairs, (2) near-core network contains the next highest represented 1-5% of the gene-pairs, (3) private-network contains ~ 50% of the gene pairs that are unique to individual genomes, and (4) the dispensable-network contains the remaining 50-95% of the gene-pairs in the maize pan-genome. Strikingly, the private-network contained almost all the genes in the pan-network but lacked half of the interactions. We performed gene ontology (GO) enrichment analysis for the pan-, core-, and private- networks and compared the contributions of variants overlapping with genes and promoters to the GWAS-driven pan-network.

Conclusions

Gene co-expression networks revealed meaningful information about groups of co-regulated genes that play a central role in regulatory processes. Pan-network approach enabled us to visualize the global view of the gene regulatory network for the studied system that could not be well inferred by the core-network alone.

Of different types of noncoding RNAs, microRNAs (miRNAs) have arguably been in the spotlight over the last decade. As post-transcriptional regulators of gene expression, miRNAs play key roles in various cellular pathways, including both development and response to a/biotic stress, such as drought and diseases. Having high-quality reference genome sequences enabled identification and annotation of miRNAs in several plant species, where miRNA sequences are highly conserved. As computational miRNA identification and annotation processes are mostly error-prone processes, homology-based predictions increase prediction accuracy. We developed and have improved the miRNA annotation pipeline, SUmir, in the last decade, which has been used for several plant genomes since then. This study presents a fully automated, new miRNA pipeline, mirMachine (miRNA Machine), by (i) adding an additional filtering step on the secondary structure predictions, (ii) making it fully automated, and (iii) introducing new options to predict either known miRNA based on homology or novel miRNAs based on small RNA sequencing reads using the previous pipeline. The new miRNA pipeline, mirMachine, was tested using The Arabidopsis Information Resource, TAIR10, release of the Arabidopsis genome and the International Wheat Genome Sequencing Consortium (IWGSC) wheat reference genome v2.

Following the elucidation of the critical roles they play in numerous important biological processes, long noncoding RNAs (lncRNAs) have gained vast attention in recent years. Manual annotation of lncRNAs is restricted by known gene annotations and is prone to false prediction due to the incompleteness of available data. However, with the advent of high-throughput sequencing technologies, a magnitude of high-quality data has become available for annotation, especially for plant species such as wheat. Here, we compared prediction accuracies of several machine learning algorithms using a 10-fold cross-validation. This study includes a comprehensive feature selection step to refine irrelevant and repeated features. We present a crop-specific, alignment-free coding potential prediction tool, LncMachine, that performs at higher prediction accuracies than the currently available popular tools (CPC2, CPAT, and CNIT) when used with the Random Forest algorithm. Further, LncMachine with Random Forest performed well on human and mouse data, with an average accuracy of 92.67%. LncMachine only requires either a FASTA file or a TAB separated CSV file containing features as input files. LncMachine can deploy several user-provided algorithms in real time and therefore be effortlessly applied to a wide range of studies.

Curated resources at centralized repositories provide high-value service to users by enhancing data veracity. Curation, however, comes with a cost, as it requires dedicated time and effort from personnel with deep domain knowledge. In this paper, we investigate the performance of a large language model (LLM), specifically generative pre-trained transformer (GPT)-3.5 and GPT-4, in extracting and presenting data against a human curator. In order to accomplish this task, we used a small set of journal articles on wheat and barley genetics, focusing on traits, such as salinity tolerance and disease resistance, which are becoming more important. The 36 papers were then curated by a professional curator for the GrainGenes database (https://wheat.pw.usda.gov). In parallel, we developed a GPT-based retrieval-augmented generation question-answering system and compared how GPT performed in answering questions about traits and quantitative trait loci (QTLs). Our findings show that on average GPT-4 correctly categorized manuscripts 97% of the time, correctly extracted 80% of traits, and 61% of marker-trait associations. Furthermore, we assessed the ability of a GPT-based DataFrame agent to filter and summarize curated wheat genetics data, showing the potential of human and computational curators working side-by-side. In one case study, our findings show that GPT-4 was able to retrieve up to 91% of disease related, human-curated QTLs across the whole genome, and up to 96% across a specific genomic region through prompt engineering. Also, we observed that across most tasks, GPT-4 consistently outperformed GPT-3.5 while generating less hallucinations, suggesting that improvements in LLM models will make generative artificial intelligence a much more accurate companion for curators in extracting information from scientific literature. Despite their limitations, LLMs demonstrated a potential to extract and present information to curators and users of biological databases, as long as users are aware of potential inaccuracies and the possibility of incomplete information extraction.