Reproducing a published figure: Tn7-like transposons • geneviewer

Intro

This tutorial shows how to rebuild a figure from a paper with geneviewer. As an example we redraw Figure 3 of Peters JE, Makarova KS, Shmakov S, Koonin EV (2017). Recruitment of CRISPR-Cas systems by Tn7-like transposons. Proc Natl Acad Sci USA 114(35):E7358-E7366, doi:10.1073/pnas.1709035114. The figure compares Tn7, Tn6022 and fourteen Tn7-like transposons that carry a minimal CRISPR-Cas system. Each element is drawn from its right end, next to the gene it inserted beside, and the elements are lined up on that end.

Open Figure 3 next to this tutorial: at each step we point out which part of the published figure we recreate.

Step by step we will line the elements up, leave out long stretches of DNA with scale breaks, draw the ends of the elements as bars, write the gene names inside the genes and finish the chart for a publication. Unlike the published figure, the genes are drawn to scale. Several of the options used here (groupStyle, style_by and a scale_break_padding in base pairs) are only available in the development version of geneviewer.

Materials

The data comes with geneviewer as the tn7_crispr_elements data set. It holds the genes of the sixteen elements of the figure, read from their GenBank records with read_gbk(), together with the positions of the ends of each element. The GenBank files of the sixteen regions, downloaded from GenBank (the accessions are in the accession column), and the data set as a .csv file can be downloaded from the geneviewer-tutorials repository. The script that made the data set from the GenBank files, data-raw/tn7_crispr_elements.R, is in the geneviewer repository.

cluster	strain	gene	label	product	category	start	end	strand
Tn7	Acinetobacter baumannii UC20804	attTn7 (glmS)	glmS	glutamine–fructose-6-phosphate transaminase (isomerizing)	insertion site	28	1866	forward
Tn7	Acinetobacter baumannii UC20804	R		right end of the element	right end	1823	1823	forward
Tn7	Acinetobacter baumannii UC20804	tnsA	tnsA	TnsA endonuclease N-terminal domain-containing protein	tnsA	2026	2847	forward
Tn7	Acinetobacter baumannii UC20804	tnsB	tnsB	transposase family protein	tnsB	2834	4942	forward
Tn7	Acinetobacter baumannii UC20804	tnsC	tnsC	AAA family ATPase	tnsC	4939	6606	forward
Tn7	Acinetobacter baumannii UC20804	tniQ	tniQ	TnsD family transposase	tniQ / tnsD	6609	8135	forward
Tn7	Acinetobacter baumannii UC20804	tnsE	tnsE	transcriptional antiterminator	tnsE	8136	9752	forward
Tn7	Acinetobacter baumannii UC20804			IS256 family transposase	other	9808	9960	complement

Each element is a cluster. The category column holds the role of each gene and sets its colour: the transposition genes (tnsA to tnsE), the cas genes, the CRISPR arrays, the insertion site and “other” genes. The label column holds the names the published figure writes on the genes. Besides the genes, each element has a row for its right end (gene R, category “right end”) and its left end (gene L, category “left end”), each a single position.

A first chart

In Figure 3: each row is one element, and the genes are coloured by their role: red tnsA, blue tnsB, light blue tnsC, green tniQ, dark red, pink and purple for the cas genes, and grey for the other genes. The figure has no legend; the caption explains the colours, and we will add a legend to our chart.

The ends of the elements are single positions, so they would have no width in the chart. We give each end a width of 100 bp, from the end into the element, so it stays clear of the gene next to it.

genes <- tn7_crispr_elements %>%
  mutate(end = ifelse(gene == "R", end + 100, end),
         start = ifelse(gene == "L", start - 100, start))

We load the genes into GC_chart(), colour them by category with GC_color() and set the order of the legend. With axis_type = "range" every element starts at 1, so the scale shows lengths rather than positions on the genome; we hide it, as the published figure has none.

colors <- list(
  "insertion site" = "#F3C79B",
  "tnsA" = "#D62728",
  "tnsB" = "#AEB6F0",
  "tnsC" = "#D7F2F2",
  "tniQ / tnsD" = "#2CA02C",
  "tnsE" = "#C7E02C",
  "cas8f / cas5f" = "#8C1515",
  "cas7f" = "#F2A5C4",
  "cas6f" = "#7B3FD4",
  "cas5" = "#B5651D",
  "cas, other" = "#C9A0DC",
  "CRISPR array" = "#555555",
  "right end" = "#9FD4DE",
  "left end" = "#F28E2B",
  "other" = "#E8E8E8"
)

chart <- GC_chart(genes, cluster = "cluster", group = "category",
                  strand = "strand", height = "900px") %>%
  GC_color(customColors = colors) %>%
  GC_scale(axis_type = "range", hidden = TRUE) %>%
  GC_legend(position = "bottom", align = "center", order = names(colors))

chart

Most elements run on far beyond the genes shown: the left end of element #438 lies over 100 kb further on. All elements share one scale, so these long stretches squeeze the genes of every row together.

Leaving out long stretches

In Figure 3: the long elements end with // on the line just before their left end. The figure leaves out tens of kilobases of cargo there, so every element fits on one row.

Scale breaks leave out a stretch of DNA and mark the gap with // on the line. With scale_breaks = TRUE, GC_scale() puts a break in every gap that is longer than scale_break_threshold percent (20% by default) of a row. scale_break_padding sets how much of the gap is kept on either side of the break: as a percentage, or as here in base pairs, so every break gets the same room.

chart <- chart %>%
  GC_scale(axis_type = "range", hidden = TRUE, scale_breaks = TRUE,
           scale_break_padding = "500bp")

chart

Lining up the elements

In Figure 3: the light blue box marked “R” is in the same column in every row, with tnsA right after it. The gene the element inserted beside is to its left.

In the published figure the elements are lined up on their right end. GC_align() does this: it shifts each row so that the gene we choose, here the right end R, sits at the same place in every row. All rows keep the same scale, and the scale breaks are left out when the rows are lined up.

chart <- chart %>%
  GC_align(id_column = "gene", id = "R", align = "left")

chart

Drawing the ends and the CRISPR arrays

In Figure 3: the ends of each element are small light blue boxes with “R” and “L” in them, and the CRISPR arrays are black stripes next to cas6. The figure is not to scale; in our to-scale chart the 100 bp ends are too narrow for a letter, so we tell them apart by colour and the legend instead.

So far every gene is drawn as an arrow, the ends of the elements too. With the groupStyle option of GC_genes() we can change the shape of all genes of a group at once. By default the groups are those that set the colours (the category column), but style_by lets us use another column: here the gene column, where the ends are R and L, the CRISPR arrays CRISPR and the fused tnsA of element #2757 tnsA (fused). We draw the ends as tall bars, and the CRISPR arrays and the fused tnsA as boxes the height of the genes. The two ends also get a diagonal pattern with GC_pattern(), on top of their colours (background = NULL); groups left out of the list get no pattern.

end_bar <- list(marker = "box", markerHeight = 38, arrowheadHeight = 38)
gene_box <- list(marker = "box", markerHeight = 20, arrowheadHeight = 20)
shapes <- list(
  R = end_bar,
  L = end_bar,
  CRISPR = c(gene_box, stroke = "black", strokeWidth = 1),
  "tnsA (fused)" = gene_box
)

chart <- chart %>%
  GC_genes(marker_size = "small", markerHeight = 20,
           style_by = "gene", groupStyle = shapes) %>%
  GC_pattern(patterns = list("right end" = "diagonal", "left end" = "diagonal"),
             background = NULL, size = 4)

chart

Labels inside the genes

In Figure 3: the gene names are written inside the arrows, including the name of the insertion site in the first gene, e.g. glmS or SRP-RNA. Only the transposition genes, the cas genes, the insertion sites and a few cargo genes of Tn7 are named; the other genes have no name.

The published figure writes the gene names inside the genes. We take the names from the label column and move them from above the genes into their middle with y and dy. adjustLabels = FALSE keeps them straight, rather than rotating labels that overlap.

Labels can be styled by group too. With style_by = "label" the groups are the label texts themselves, so we can make the names on the dark genes white and move the label of the SRP RNA genes, which are too short for it, above the gene and to the left of the right end beside it.

white <- list(fill = "white")
label_styles <- list(
  tnsA = white, cas8f = white, cas6f = white, cas5 = white,
  "SRP-RNA" = list(y = 55, dy = "0px", textAnchor = "end", x = -4)
)

chart <- chart %>%
  GC_labels(label = "label", fontSize = "9px", y = 50, dy = "14.5px",
            adjustLabels = FALSE, style_by = "label", groupStyle = label_styles)

chart

Finishing the chart

In Figure 3: each row is named on the left with its node in the TnsA tree (#2, #141 and so on). We add the host of each element below its name, and a title that cites the figure.

To name the rows we take the name of each element and its host. The strain_short column holds the host with its genus abbreviated, e.g. V. cholerae VC35, so the labels stay short.

elements <- distinct(genes, cluster, strain_short)

To finish the chart we name each row with its element and host using GC_clusterLabel(), add a title that cites the published figure, show the product of each gene in the tooltip and apply the "publication" theme. In the legend, groupStyle sets the gene names in italics, while the other entries stay upright.

gene_names <- c("tnsA", "tnsB", "tnsC", "tniQ / tnsD", "tnsE", "cas8f / cas5f",
                "cas7f", "cas6f", "cas5")
legend_styles <- lapply(setNames(nm = gene_names),
                        function(x) list(fontStyle = "italic"))

chart <- chart %>%
  GC_clusterLabel(
    title = elements$cluster,
    subtitle = elements$strain_short,
    width = "190px",
    fontSize = "12px",
    subtitleFont = list(fontSize = "10px", fontStyle = "italic"),
    align = "right",
    x = -10
  ) %>%
  GC_title(
    "Tn7-like transposons that carry a minimal CRISPR-Cas system",
    subtitle = paste("Redrawn from Fig. 3 of Peters et al. (2017),",
                     "Proc Natl Acad Sci USA 114:E7358-E7366"),
    height = "70px",
    titleFont = list(fontSize = "19px", fontWeight = "bold"),
    subtitleFont = list(fontSize = "14px", fontStyle = "italic")
  ) %>%
  GC_tooltip(formatter = "<b>{gene}</b> {product}<br>{start} - {end}") %>%
  GC_legend(legendTextOptions = list(fontSize = "12px"),
            groupStyle = legend_styles) %>%
  GC_theme("publication")

chart

Saving the chart

To save the chart for a publication, hover over the download icon at the top right of the chart and choose “Save as SVG”. The SVG holds the whole chart, with all rows, the title and the legend, as one vector file, so it stays sharp at any size. It can also be opened in a vector editor such as Inkscape, Adobe Illustrator or Affinity Designer to finish the figure by hand, for example to move a label, add an arrow or combine it with other panels. Choose “Save as PNG” for a picture to use in slides or on the web.

To change the look of the chart further, for example other colours, fonts or marker shapes, see Get started or the documentation of each function, e.g. ?GC_genes.