Table 1.
Large-scale Orthopteran transcriptome resources to date.
Figure 1.
Oogenesis and embryogenesis in the cricket model organism Gryllus bimaculatus.
(A) Adult female cricket perched on a gloved human finger for perspective. (B) Anterior tip of a single ovariole from an adult female ovary, showing oocytes (o) at early previtellogenic stages of oogenesis. A single large germinal vesicle (gv) is distinguishable in each oocyte. Unlike meroistic (containing nurse cells) Drosophila ovaries, G. bimaculatus ovaries are panoistic and lack nurse cells [100]. (C) A single late stage oocyte with a single layer of columnar follicle cells (fc). (D–J) Chronological stages of G. bimaculatus embryogenesis showing the range of embryonic stages represented in the transcriptome presented here. (D) A fertilized egg just after laying. The egg nucleus is distinguishable as a dense patch in the dorsal yolk (arrowhead). Ages are shown as days (d) after egg-laying at 29°C. (E–I) are 3D reconstructions of confocal optical sections of Hoechst 3342-stained embryos dissected free from the egg; (J) is a micrograph of a live embryo dissected free from the chorion. Abbreviations: A = abdomen; C = cerci; E = eye; H = head; G = gnathal segments; L1 = first thoracic leg; L2 = second thoracic leg; L3 = third thoracic leg; T = thorax. Scale bar is 100 µm in (B, C, E–I) and 500 µm in (D, J). Anterior is to the left in all panels. Photo in (A) courtesy of David Behl; photos in (D) and (J) from [101].
Table 2.
Assembly statistics and BLAST results against nr for the G. bimaculatus de novo transcriptome assembly.
Figure 2.
Distribution of average coverage (bp/contig) within contigs produced by de novo assembly of the G. bimaculatus transcriptome.
The coverage within contigs is calculated by dividing the total number of base pairs contained in the reads used to construct a contig by the length of that contig.
Figure 3.
Assessment of gene discovery and read length capacity of the G. bimaculatus de novo assembled transcriptome.
(A) Randomly selected subsets of the trimmed reads were assembled using Newbler v2.5 in 10% increments, up to and including 100% of trimmed reads. For each subassembly, the number of unique BLAST hits against the NCBI non-redundant database (nr) with an E-value cutoff of 1e-10 (red; left axis) and the average coverage per base pair (blue; right axis) was calculated (see text for details). The number of unique BLAST hits did not increase after at least 90% of reads (3,795,085 reads) were assembled, while the coverage per base pair continued to increase as reads were added to the assembly. (B) Isotig length distribution for each subassembly created as described in (A). (C) Isotig length distribution of each subassembly for isotigs ≥4 kb. High numbers (≥50) of isotigs over 4 kb in length are achieved only when ≥40% of reads (1,686,646 reads) are assembled.
Figure 4.
Ortholog hit ratio analysis of the G. bimaculatus de novo assembled transcriptome.
The ortholog hit ratio is a comparison of the length of an assembled sequence to the total length of the full length transcript of its putative ortholog [71]. Values close to one suggest that a transcript predicted by the de novo assembly is close to full length. Ortholog hit ratios for the G. bimaculatus transcriptome sequences are compared to those for the previously reported de novo assembled transcriptome of another insect, the milkweed bug Oncopeltus fasciatus [11]. (A) Ortholog hit ratio analysis of assembled isotigs. A majority (63.8%) of all G. bimaculatus isotigs (black bars) have an ortholog hit ratio of ≥0.5 (blue arrowhead), and 40.0% have an ortholog hit ratio of ≥0.8 (red arrowhead). These values are higher than those obtained for the O. fasciatus de novo assembled transcriptome (grey bars) [11]. (B) Ortholog hit ratio analysis of unassembled singletons. As expected, singletons represent much smaller proportions of putative full-length transcripts. 6.3% of G. bimaculatus singletons (black) have an ortholog hit ratio of ≥0.5 (blue arrowhead), while 0.8% have an ortholog hit ratio of ≥0.8 (red arrowhead). As for the isotig analysis, these values are higher than those obtained for the O. fasciatus de novo assembled transcriptome (grey) [11].
Figure 5.
Phylogenetic comparison of proportion of known proteomes represented in the G. bimaculatus de novo assembled transcriptome.
The number (bold) and percentage (bold italics) of proteome sequences with a putative G. bimaculatus ortholog in the de novo transcriptome assembly is shown for selected animals with sequenced genomes (based on top BLAST hit, E-value cutoff 1e-5). Proteomes were predicted from genome sequence sources as shown in Table S1. Numbers in large font in red and blue ovals indicate average proportion of sequences from all tested insect and deuterostome proteomes, respectively, represented in the G. bimaculatus transcriptome.
Figure 6.
Sequence extension and gene discovery in the G. bimaculatus Hedgehog and Hippo pathways.
(A) The de novo transcriptome assembly of G. bimaculatus newly identifies most members of the hedgehog pathway (red), from which only the hedgehog ligand (blue) was previously known (GenBank accession AB044709). (B) The transcriptome also adds significant sequence data to the fragments of many genes in the Hippo signaling pathway that had been previously identified (green). Seven genes of the known pathway were not identified in the transcriptome (yellow, white), two of which lack any sequence data in GenBank (white). GenBank accessions for previously identified sequences are as follows: discs overgrown (dco): AB443442; expanded (ex): AB378099; warts (wts): AB300574; cyclin E (cycE): AB378067; hippo (hpo): AB378070; inhibitor of apoptosis protein (diap1): AB378071; mob as tumor suppressor (mats): AB378072; yorkie (yki): AB378076; scaffold protein salvador (sav): AB378074; Merlin (Mer): AB378073; Kibra: DC445461.
Figure 7.
Automated annotation of the G. bimaculatus de novo transcriptome assembly using Gene Predictor.
(A) Comparison of the proportion of non-redundant assembly sequences, isotigs and singletons that obtained a significant BLAST hit against nr (black bars), and those that were assigned a putative orthology by Gene Predictor (GP; white bars), based on the best reciprocal top BLAST hit with the Drosophila melanogaster proteome (see Table S1). (B) Comparison of the proportion of sequences with a significant BLAST hit in nr that also had a putative orthology assignment based on Gene Predictor (dark grey bars). All sequences assigned putative orthologs by Gene Predictor also had significant BLAST hits in nr (light grey bars).
Figure 8.
Coding region analysis of G. bimaculatus de novo transcriptome assembly sequences without significant BLAST hits in nr.
Assembly products that failed to obtain significant BLAST hits in nr (white) were examined for the presence of coding regions (green) using EST Scan [52]. Assembly sequences thus predicted to contain coding regions were examined for the presence of known coding domains (yellow) using InterPro Scan [53], [54]. Results are shown separately for isotigs (A), singletons (B) and all non-redundant assembly products (C). See also Table 3.
Table 3.
Length parameters of isotigs according to BLAST annotation and predicted protein-coding status.
Table 4.
Length parameters of singletons according to BLAST annotation and predicted protein-coding status.
Table 5.
Statistical comparison of isotig and singleton nucleotide sequence lengths according to BLAST annotation and predicted protein-coding status.
Table 6.
Statistical comparison of isotig and singleton predicted coding sequence lengths according to BLAST annotation status.
Figure 9.
Comparison of sequences lacking significant BLAST hits to nr, with Laupala kohalensis and Locusta migratoria databases.
(A–C) Assembly products that failed to obtain significant BLAST hits to nr (white) were examined for significant similarity (magenta) to transcripts from at least one of L. migratoria or L. kohalensis [72], [73], [74], [75]. (A′–C′) Assembly sequences thus identified were parsed into sequences with significant hits among only L. kohalensis sequences (red), only L. migratoria sequences (blue), or both (yellow). Results are shown separately for isotigs (A, A′), singletons (B, A′) and all non-redundant assembly products (C, A′).
Figure 10.
Principal protein domain composition of G. bimaculatus transcriptome sequences with highest similarity to Laupala kohalensis or Locusta migratoria sequences.
Relative proportions of the top 25 protein domains coded by G. bimaculatus transcriptome sequences with significant similarity to sequences from L. kohalensis (A), L. migratoria (B), or sequences from nr (C). Protein domain nomenclature from Pfam [102] as follows: AdoHcyase_NAD: PF00670; Ank: PF00023; ATP-gua_Ptrans/N: PF02807; BTB/POZ: PF00651; C2: PF00168; DUF (combined): n/a; EFG domains (combined): n/a; efhand/like: PF09279; F-box: PF00646; Glyco_hydro (combined): n/a; GTP_EFTU domains: PF00009; Laps: PF10169; LRR_1: PF00560; Metallophos: PF00149; Myb_DNA-binding (combined): n/a; OS-D: PF03392; PARP: PF00644; PGAMP: PF07644; Pkinase: PF00069; Ras: PF00071; Ribosomal (combined): n/a; RRM_1: PF00076; RVT_1: PF00078; ubiquitin: PF00240; zinc finger (combined): n/a. “Combined” indicates that multiple Pfam accessions are combined.