Consideration of non-canonical splice sites improves gene prediction on the Arabidopsis thaliana Niederzenz-1 genome sequence

DRANetSplicer: A Splice Site Prediction Model Based on Deep Residual Attention Networks

The precise identification of splice sites is essential for unraveling the structure and function of genes, constituting a pivotal step in the gene annotation process. In this study, we developed a novel deep learning model, DRANetSplicer, that integrates residual learning and attention mechanisms for enhanced accuracy in capturing the intricate features of splice sites. We constructed multiple datasets using the most recent versions of genomic data from three different organisms, Oryza sativa japonica, Arabidopsis thaliana and Homo sapiens. This approach allows us to train models with a richer set of high-quality data. DRANetSplicer outperformed benchmark methods on donor and acceptor splice site datasets, achieving an average accuracy of (96.57%, 95.82%) across the three organisms. Comparative analyses with benchmark methods, including SpliceFinder, Splice2Deep, Deep Splicer, EnsembleSplice, and DNABERT, revealed DRANetSplicer's superior predictive performance, resulting in at least a (4.2%, 11.6%) relative reduction in average error rate. We utilized the DRANetSplicer model trained on O. sativa japonica data to predict splice sites in A. thaliana, achieving accuracies for donor and acceptor sites of (94.89%, 94.25%). These results indicate that DRANetSplicer possesses excellent cross-organism predictive capabilities, with its performance in cross-organism predictions even surpassing that of benchmark methods in non-cross-organism predictions. Cross-organism validation showcased DRANetSplicer's excellence in predicting splice sites across similar organisms, supporting its applicability in gene annotation for understudied organisms. We employed multiple methods to visualize the decision-making process of the model. The visualization results indicate that DRANetSplicer can learn and interpret well-known biological features, further validating its overall performance. Our study systematically examined and confirmed the predictive ability of DRANetSplicer from various levels and perspectives, indicating that its practical application in gene annotation is justified.

Identification of annotation artifacts concerning the chalcone synthase (CHS)

OBJECTIVE

Chalcone synthase (CHS) catalyzes the initial step of the flavonoid biosynthesis. The CHS encoding gene is well studied in numerous plant species. Rapidly growing sequence databases contain hundreds of CHS entries that are the result of automatic annotation. In this study, we evaluated apparent multiplication of CHS domains in CHS gene models of four plant species.

MAIN FINDINGS

CHS genes with an apparent triplication of the CHS domain encoding part were discovered through database searches. Such genes were found in Macadamia integrifolia, Musa balbisiana, Musa troglodytarum, and Nymphaea colorata. A manual inspection of the CHS gene models in these four species with massive RNA-seq data suggests that these gene models are the result of artificial fusions in the annotation process. While there are hundreds of seemingly correct CHS records in the databases, it is not clear why these annotation artifacts appeared.

Apiaceae FNS I originated from F3H through tandem gene duplication

BACKGROUND

Flavonoids are specialized metabolites with numerous biological functions in stress response and reproduction of plants. Flavones are one subgroup that is produced by the flavone synthase (FNS). Two distinct enzyme families evolved that can catalyze the biosynthesis of flavones. While the membrane-bound FNS II is widely distributed in seed plants, one lineage of soluble FNS I appeared to be unique to Apiaceae species.

RESULTS

We show through phylogenetic and comparative genomic analyses that Apiaceae FNS I evolved through tandem gene duplication of flavanone 3-hydroxylase (F3H) followed by neofunctionalization. Currently available datasets suggest that this event happened within the Apiaceae in a common ancestor of Daucus carota and Apium graveolens. The results also support previous findings that FNS I in the Apiaceae evolved independent of FNS I in other plant species.

CONCLUSION

We validated a long standing hypothesis about the evolution of Apiaceae FNS I and predicted the phylogenetic position of this event. Our results explain how an Apiaceae-specific FNS I lineage evolved and confirm independence from other FNS I lineages reported in non-Apiaceae species.

Mapping‑by‑Sequencing Reveals Genomic Regions Associated with Seed Quality Parameters in Brassica napus

Rapeseed (Brassica napus L.) is an important oil crop and has the potential to serve as a highly productive source of protein. This protein exhibits an excellent amino acid composition and has high nutritional value for humans. Seed protein content (SPC) and seed oil content (SOC) are two complex quantitative and polygenic traits which are negatively correlated and assumed to be controlled by additive and epistatic effects. A reduction in seed glucosinolate (GSL) content is desired as GSLs cause a stringent and bitter taste. The goal here was the identification of genomic intervals relevant for seed GSL content and SPC/SOC. Mapping by sequencing (MBS) revealed 30 and 15 new and known genomic intervals associated with seed GSL content and SPC/SOC, respectively. Within these intervals, we identified known but also so far unknown putatively causal genes and sequence variants. A 4 bp insertion in the MYB28 homolog on C09 shows a significant association with a reduction in seed GSL content. This study provides insights into the genetic architecture and potential mechanisms underlying seed quality traits, which will enhance future breeding approaches in B. napus.

Spliceator: multi-species splice site prediction using convolutional neural networks

Background

Ab initio prediction of splice sites is an essential step in eukaryotic genome annotation. Recent predictors have exploited Deep Learning algorithms and reliable gene structures from model organisms. However, Deep Learning methods for non-model organisms are lacking.

Results

We developed Spliceator to predict splice sites in a wide range of species, including model and non-model organisms. Spliceator uses a convolutional neural network and is trained on carefully validated data from over 100 organisms. We show that Spliceator achieves consistently high accuracy (89–92%) compared to existing methods on independent benchmarks from human, fish, fly, worm, plant and protist organisms.

Conclusions

Spliceator is a new Deep Learning method trained on high-quality data, which can be used to predict splice sites in diverse organisms, ranging from human to protists, with consistently high accuracy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-021-04471-3.

Characterization of the Brassica napus Flavonol Synthase Gene Family Reveals Bifunctional Flavonol Synthases

Flavonol synthase (FLS) is a key enzyme for the formation of flavonols, which are a subclass of the flavonoids. FLS catalyzes the conversion of dihydroflavonols to flavonols. The enzyme belongs to the 2-oxoglutarate-dependent dioxygenases (2-ODD) superfamily. We characterized the FLS gene family of Brassica napus that covers 13 genes, based on the genome sequence of the B. napus cultivar Express 617. The goal was to unravel which BnaFLS genes are relevant for seed flavonol accumulation in the amphidiploid species B. napus. Two BnaFLS1 homeologs were identified and shown to encode bifunctional enzymes. Both exhibit FLS activity as well as flavanone 3-hydroxylase (F3H) activity, which was demonstrated in vivo and in planta. BnaFLS1-1 and -2 are capable of converting flavanones into dihydroflavonols and further into flavonols. Analysis of spatio-temporal transcription patterns revealed similar expression profiles of BnaFLS1 genes. Both are mainly expressed in reproductive organs and co-expressed with the genes encoding early steps of flavonoid biosynthesis. Our results provide novel insights into flavonol biosynthesis in B. napus and contribute information for breeding targets with the aim to modify the flavonol content in rapeseed.

Rapid protein evolution, organellar reductions, and invasive intronic elements in the marine aerobic parasite dinoflagellate Amoebophrya spp

BACKGROUND

Dinoflagellates are aquatic protists particularly widespread in the oceans worldwide. Some are responsible for toxic blooms while others live in symbiotic relationships, either as mutualistic symbionts in corals or as parasites infecting other protists and animals. Dinoflagellates harbor atypically large genomes (~ 3 to 250 Gb), with gene organization and gene expression patterns very different from closely related apicomplexan parasites. Here we sequenced and analyzed the genomes of two early-diverging and co-occurring parasitic dinoflagellate Amoebophrya strains, to shed light on the emergence of such atypical genomic features, dinoflagellate evolution, and host specialization.

RESULTS

We sequenced, assembled, and annotated high-quality genomes for two Amoebophrya strains (A25 and A120), using a combination of Illumina paired-end short-read and Oxford Nanopore Technology (ONT) MinION long-read sequencing approaches. We found a small number of transposable elements, along with short introns and intergenic regions, and a limited number of gene families, together contribute to the compactness of the Amoebophrya genomes, a feature potentially linked with parasitism. While the majority of Amoebophrya proteins (63.7% of A25 and 59.3% of A120) had no functional assignment, we found many orthologs shared with Dinophyceae. Our analyses revealed a strong tendency for genes encoded by unidirectional clusters and high levels of synteny conservation between the two genomes despite low interspecific protein sequence similarity, suggesting rapid protein evolution. Most strikingly, we identified a large portion of non-canonical introns, including repeated introns, displaying a broad variability of associated splicing motifs never observed among eukaryotes. Those introner elements appear to have the capacity to spread over their respective genomes in a manner similar to transposable elements. Finally, we confirmed the reduction of organelles observed in Amoebophrya spp., i.e., loss of the plastid, potential loss of a mitochondrial genome and functions.

CONCLUSION

These results expand the range of atypical genome features found in basal dinoflagellates and raise questions regarding speciation and the evolutionary mechanisms at play while parastitism was selected for in this particular unicellular lineage.

The reuse of public datasets in the life sciences: potential risks and rewards

The 'big data' revolution has enabled novel types of analyses in the life sciences, facilitated by public sharing and reuse of datasets. Here, we review the prodigious potential of reusing publicly available datasets and the associated challenges, limitations and risks. Possible solutions to issues and research integrity considerations are also discussed. Due to the prominence, abundance and wide distribution of sequencing data, we focus on the reuse of publicly available sequence datasets. We define 'successful reuse' as the use of previously published data to enable novel scientific findings. By using selected examples of successful reuse from different disciplines, we illustrate the enormous potential of the practice, while acknowledging the respective limitations and risks. A checklist to determine the reuse value and potential of a particular dataset is also provided. The open discussion of data reuse and the establishment of this practice as a norm has the potential to benefit all stakeholders in the life sciences.

High Contiguity De Novo Genome Sequence Assembly of Trifoliate Yam (Dioscorea dumetorum) Using Long Read Sequencing

Trifoliate yam (Dioscorea dumetorum) is one example of an orphan crop, not traded internationally. Post-harvest hardening of the tubers of this species starts within 24 h after harvesting and renders the tubers inedible. Genomic resources are required for D. dumetorum to improve breeding for non-hardening varieties as well as for other traits. We sequenced the D. dumetorum genome and generated the corresponding annotation. The two haplophases of this highly heterozygous genome were separated to a large extent. The assembly represents 485 Mbp of the genome with an N50 of over 3.2 Mbp. A total of 35,269 protein-encoding gene models as well as 9941 non-coding RNA genes were predicted, and functional annotations were assigned.

Animal, Fungi, and Plant Genome Sequences Harbor Different Non-Canonical Splice Sites

Most protein-encoding genes in eukaryotes contain introns, which are interwoven with exons. Introns need to be removed from initial transcripts in order to generate the final messenger RNA (mRNA), which can be translated into an amino acid sequence. Precise excision of introns by the spliceosome requires conserved dinucleotides, which mark the splice sites. However, there are variations of the highly conserved combination of GT at the 5' end and AG at the 3' end of an intron in the genome. GC-AG and AT-AC are two major non-canonical splice site combinations, which have been known for years. Recently, various minor non-canonical splice site combinations were detected with numerous dinucleotide permutations. Here, we expand systematic investigations of non-canonical splice site combinations in plants across eukaryotes by analyzing fungal and animal genome sequences. Comparisons of splice site combinations between these three kingdoms revealed several differences, such as an apparently increased CT-AC frequency in fungal genome sequences. Canonical GT-AG splice site combinations in antisense transcripts are a likely explanation for this observation, thus indicating annotation errors. In addition, high numbers of GA-AG splice site combinations were observed in Eurytemoraaffinis and Oikopleuradioica. A variant in one U1 small nuclear RNA (snRNA) isoform might allow the recognition of GA as a 5' splice site. In depth investigation of splice site usage based on RNA-Seq read mappings indicates a generally higher flexibility of the 3' splice site compared to the 5' splice site across animals, fungi, and plants.

A chromosome-level sequence assembly reveals the structure of the Arabidopsis thaliana Nd-1 genome and its gene set

In addition to the BAC-based reference sequence of the accession Columbia-0 from the year 2000, several short read assemblies of THE plant model organism Arabidopsis thaliana were published during the last years. Also, a SMRT-based assembly of Landsberg erecta has been generated that identified translocation and inversion polymorphisms between two genotypes of the species. Here we provide a chromosome-arm level assembly of the A. thaliana accession Niederzenz-1 (AthNd-1_v2c) based on SMRT sequencing data. The best assembly comprises 69 nucleome sequences and displays a contig length of up to 16 Mbp. Compared to an earlier Illumina short read-based NGS assembly (AthNd-1_v1), a 75 fold increase in contiguity was observed for AthNd-1_v2c. To assign contig locations independent from the Col-0 gold standard reference sequence, we used genetic anchoring to generate a de novo assembly. In addition, we assembled the chondrome and plastome sequences. Detailed analyses of AthNd-1_v2c allowed reliable identification of large genomic rearrangements between A. thaliana accessions contributing to differences in the gene sets that distinguish the genotypes. One of the differences detected identified a gene that is lacking from the Col-0 gold standard sequence. This de novo assembly extends the known proportion of the A. thaliana pan-genome.

Genome-wide analyses supported by RNA-Seq reveal non-canonical splice sites in plant genomes

BACKGROUND

Most eukaryotic genes comprise exons and introns thus requiring the precise removal of introns from pre-mRNAs to enable protein biosynthesis. U2 and U12 spliceosomes catalyze this step by recognizing motifs on the transcript in order to remove the introns. A process which is dependent on precise definition of exon-intron borders by splice sites, which are consequently highly conserved across species. Only very few combinations of terminal dinucleotides are frequently observed at intron ends, dominated by the canonical GT-AG splice sites on the DNA level.

RESULTS

Here we investigate the occurrence of diverse combinations of dinucleotides at predicted splice sites. Analyzing 121 plant genome sequences based on their annotation revealed strong splice site conservation across species, annotation errors, and true biological divergence from canonical splice sites. The frequency of non-canonical splice sites clearly correlates with their divergence from canonical ones indicating either an accumulation of probably neutral mutations, or evolution towards canonical splice sites. Strong conservation across multiple species and non-random accumulation of substitutions in splice sites indicate a functional relevance of non-canonical splice sites. The average composition of splice sites across all investigated species is 98.7% for GT-AG, 1.2% for GC-AG, 0.06% for AT-AC, and 0.09% for minor non-canonical splice sites. RNA-Seq data sets of 35 species were incorporated to validate non-canonical splice site predictions through gaps in sequencing reads alignments and to demonstrate the expression of affected genes.

CONCLUSION

We conclude that bona fide non-canonical splice sites are present and appear to be functionally relevant in most plant genomes, although at low abundance.

A major QTL on chromosome C05 significantly reduces acid detergent lignin (ADL) content and increases seed oil and protein content in oilseed rape (Brassica napus L.)

A reduction in acid detergent lignin content in oilseed rape resulted in an increase in seed oil and protein content. Worldwide increasing demand for vegetable oil and protein requires continuous breeding efforts to enhance the yield of oil and protein crop species. The oil-extracted meal of oilseed rape is currently mainly used for feeding livestock, but efforts are undertaken to use the oilseed rape protein in food production. One limiting factor is the high lignin content of black-seeded oilseed rape that negatively affects digestibility and sensory quality of food products compared to soybean. Breeding attempts to develop yellow seeded oilseed rape with reduced lignin content have not yet resulted in competitive cultivars. The objective of this work was to investigate the inheritance of seed quality in a DH population derived from the cross of the high oil lines SGDH14 and cv. Express. The DH population of 139 lines was tested in field experiments in 14 environments in north-west Europe. Seeds harvested from open pollinated plants were used for extensive seed quality analysis. A molecular marker map based on the Illumina Infinium 60 K Brassica SNP chip was used to map QTL. Amongst others, one major QTL for acid detergent lignin content, explaining 81% of the phenotypic variance, was identified on chromosome C05. Lines with reduced lignin content nevertheless did not show a yellowish appearance, but showed a reduced seed hull content. The position of the QTL co-located with QTL for oil and protein content of the defatted meal with opposite additive effects, suggesting that the reduction in lignin content resulted in an increase in oil and protein content.