GeneWaltz--A new method for reducing the false positives of gene finding

Evaluating Plant Gene Models Using Machine Learning

Gene models are regions of the genome that can be transcribed into RNA and translated to proteins, or belong to a class of non-coding RNA genes. The prediction of gene models is a complex process that can be unreliable, leading to false positive annotations. To help support the calling of confident conserved gene models and minimize false positives arising during gene model prediction we have developed Truegene, a machine learning approach to classify potential low confidence gene models using 14 gene and 41 protein-based characteristics. Amino acid and nucleotide sequence-based features were calculated for conserved (high confidence) and non-conserved (low confidence) annotated genes from the published Pisum sativum Cameor genome. These features were used to train eXtreme Gradient Boost (XGBoost) classifier models to predict whether a gene model is likely to be real. The optimized models demonstrated a prediction accuracy ranging from 87% to 90% and an F-1 score of 0.91–0.94. We used SHapley Additive exPlanations (SHAP) and feature importance plots to identify the features that contribute to the model predictions, and we show that protein and gene-based features can be used to build accurate models for gene prediction that have applications in supporting future gene annotation processes.

Genomes of the willow-galling sawflies Euura lappo and Eupontania aestiva (Hymenoptera: Tenthredinidae): a resource for research on ecological speciation, adaptation, and gall induction

Hymenoptera is a hyperdiverse insect order represented by over 153,000 different species. As many hymenopteran species perform various crucial roles for our environments, such as pollination, herbivory, and parasitism, they are of high economic and ecological importance. There are 99 hymenopteran genomes in the NCBI database, yet only five are representative of the paraphyletic suborder Symphyta (sawflies, woodwasps, and horntails), while the rest represent the suborder Apocrita (bees, wasps, and ants). Here, using a combination of 10X Genomics linked-read sequencing, Oxford Nanopore long-read technology, and Illumina short-read data, we assembled the genomes of two willow-galling sawflies (Hymenoptera: Tenthredinidae: Nematinae: Euurina): the bud-galling species Euura lappo and the leaf-galling species Eupontania aestiva. The final assembly for E. lappo is 259.85 Mbp in size, with a contig N50 of 209.0 kbp and a BUSCO score of 93.5%. The E. aestiva genome is 222.23 Mbp in size, with a contig N50 of 49.7 kbp and a 90.2% complete BUSCO score. De novo annotation of repetitive elements showed that 27.45% of the genome was composed of repetitive elements in E. lappo and 16.89% in E. aestiva, which is a marked increase compared to previously published hymenopteran genomes. The genomes presented here provide a resource for inferring phylogenetic relationships among basal hymenopterans, comparative studies on host-related genomic adaptation in plant-feeding insects, and research on the mechanisms of plant manipulation by gall-inducing insects.

Genomics of ecological adaptation in cactophilic Drosophila

Cactophilic Drosophila species provide a valuable model to study gene–environment interactions and ecological adaptation. Drosophila buzzatii and Drosophila mojavensis are two cactophilic species that belong to the repleta group, but have very different geographical distributions and primary host plants. To investigate the genomic basis of ecological adaptation, we sequenced the genome and developmental transcriptome of D. buzzatii and compared its gene content with that of D. mojavensis and two other noncactophilic Drosophila species in the same subgenus. The newly sequenced D. buzzatii genome (161.5 Mb) comprises 826 scaffolds (>3 kb) and contains 13,657 annotated protein-coding genes. Using RNA sequencing data of five life-stages we found expression of 15,026 genes, 80% protein-coding genes, and 20% noncoding RNA genes. In total, we detected 1,294 genes putatively under positive selection. Interestingly, among genes under positive selection in the D. mojavensis lineage, there is an excess of genes involved in metabolism of heterocyclic compounds that are abundant in Stenocereus cacti and toxic to nonresident Drosophila species. We found 117 orphan genes in the shared D. buzzatii–D. mojavensis lineage. In addition, gene duplication analysis identified lineage-specific expanded families with functional annotations associated with proteolysis, zinc ion binding, chitin binding, sensory perception, ethanol tolerance, immunity, physiology, and reproduction. In summary, we identified genetic signatures of adaptation in the shared D. buzzatii–D. mojavensis lineage, and in the two separate D. buzzatii and D. mojavensis lineages. Many of the novel lineage-specific genomic features are promising candidates for explaining the adaptation of these species to their distinct ecological niches.

A codon substitution model that incorporates the effect of the GC contents, the gene density and the density of CpG islands of human chromosomes

Background

Developing a model for codon substitutions is essential for the analyses of protein sequences. Recent studies on the mutation rates in the non-coding regions have shown that CpG mutation rates in the human genome are negatively correlated to the local GC content and to the densities of functional elements. This study aimed at understanding the effect of genomic features, namely, GC content, gene density, and frequency of CpG islands, on the rates of codon substitution in human chromosomes.

Results

Codon substitution rates of CpG to TpG mutations, TpG to CpG mutations, and non-CpG transitions and transversions in humans were estimated by comparing the coding regions of thousands of human and chimpanzee genes and inferring their ancestral sequences by using macaque genes as the outgroup. Since the genomic features are depending on each other, partial regression coefficients of these features were obtained.

Conclusion

The substitution rates of codons depend on gene densities of the chromosomes. Transcription-associated mutation is one such pressure. On the basis of these results, a model of codon substitutions that incorporates the effect of genomic features on codon substitution in human chromosomes was developed.