Fig 4. Nonnegative matrix factorization (NMF) predicts accurate expression patterns for neural border (NB) genes. In order to represent gene expression levels on the map of the developing ectoderm, we used 2 approaches: averaging the expression values for each of 7 domains or using NMF deconvolution predictions in 5 domains. We found those 2 approaches complementary. Average patterns directly describe raw data but are sensitive to shifts in dissection. NMF deconvolution predicted expression levels in each NMF-tissue. In the case of the anterior neural border (NBa), the NMF pattern is closer to in situ hybridization patterns than average patterns, e.g., for known genes expressed in various ectoderm areas, sox2, snail2, pax3, and foxi1e (also see Fig 8, S4 and S7 Figs for more examples). Moreover, these patterns predict expression for genes of unknown pattern and with low expression level, such as sp8. We present the average pattern and NMF-predicted pattern (NMF pattern); each is normalized on a percent scale, 100% being maximal expression of the gene of interest. This implies that genes with low expression will be represented in similar shades as genes expressed at high level, which allows better visualization but could be misleading if the gene is very weakly expressed in all ectoderm (e.g., sp8 at Nieuwkoop and Faber stage [St.] 12.5). To complete pattern prediction with expression level information, we thus also include the average expression levels per dissected sample for each gene (reads per kilobase per million [RPKM] shown for St. 14 here). Robustly expressed genes (>100 RPKM, e.g., sox2, snail2, and pax3 at St. 14) are predicted as well as genes with very low expression (e.g., snail2 at St. 12.5 or sp8 at St. 14). In addition, the color of the dot indicates high ectoderm specificity (green) to low ectoderm enrichment (orange). See text and Materials and methods for details, see S11 Table for numerical data.
Image published in: Plouhinec JL et al. (2017)
© 2017 Plouhinec et al. This image is reproduced with permission of the journal and the copyright holder. This is an open-access article distributed under the terms of the Creative Commons Attribution license
| Gene | Synonyms | Species | Stage(s) | Tissue |
|---|---|---|---|---|
| snai2.L | slug, snai2-a, snai2-b, Snail2, xSlu, xslug, XSnail2 | X. laevis | Throughout NF stage 12.5 | neural plate border |
| snai2.L | slug, snai2-a, snai2-b, Snail2, xSlu, xslug, XSnail2 | X. laevis | Throughout NF stage 14 | neural plate border |
| sp8.L | btd | X. laevis | Throughout NF stage 12.5 | ectoderm |
| sp8.L | btd | X. laevis | Throughout NF stage 14 | ectoderm |
| pax3.S | cdhs, hup2, pax-3, pax3-a, pax3-b, Waardenburg syndrome 1, ws1, xpax3 | X. laevis | Throughout NF stage 12.5 | neural plate border preplacodal ectoderm |
| pax3.S | cdhs, hup2, pax-3, pax3-a, pax3-b, Waardenburg syndrome 1, ws1, xpax3 | X. laevis | Throughout NF stage 14 | neural plate border |
| foxi1.L | ectodermally-expressed mesendoderm antagonist, ema, FoxI1e, foxi1e, foxi3, Xema, xfoxi1 | X. laevis | Throughout NF stage 12.5 | non-neural ectoderm |
| foxi1.L | ectodermally-expressed mesendoderm antagonist, ema, FoxI1e, foxi1e, foxi3, Xema, xfoxi1 | X. laevis | Throughout NF stage 14 | non-neural ectoderm |
| sox2.S | anop3, LOC108718087, mcops3, Sox-2, XLSOX-2, Xsox-2, XSox2 | X. laevis | Throughout NF stage 12.5 | neural plate border neural plate |
| sox2.S | anop3, LOC108718087, mcops3, Sox-2, XLSOX-2, Xsox-2, XSox2 | X. laevis | Throughout NF stage 14 | preplacodal ectoderm neural plate neural plate border |
Image source: Published
Permanent Image Page
Printer Friendly View
XB-IMG-159578