Figure 1.
Evaluation of transposon insertion bias.
(A) Scatter plot of the genome location dependent insertion probability. Each data point represents a genomic window of 20,000 nucleotides, which was shifted by 2,000 nucleotides for each consecutive evaluation. The insertion probability gradually decreased from the origin of replication. The solid line is a best fit of ax2+b|x|+c, where a, b and c are equal to 0.0032, 0.0081 and 0.1615, respectively, and x is the location in the genome. (B) Boxplot of flanking sequence dependent insertion probability distribution showing that several motifs contained significantly more insertions than others. (C) Relative occurrence of nucleotides in the two flanking positions of inserted TA sites.
Figure 2.
Observed and Monte Carlo simulated transposon insertions.
The normalized deviation of expectation of insertions are shown for TIFA essential (red), nonessential (green) and unknown (blue) genes. The solid line represents the average outcome of 1000 Monte Carlo simulations, flanked by one standard deviation (dotted lines).
Figure 3.
(A) 50 out of 273 essential genes were associated with a fitness value.
The circumference of the circles represents the logarithmic read counts of each insertion, and the color of the circles represents the gene location of the insertion: first 10% (magenta), last 10% (cyan), and middle 80% (black). 22 essential genes contained ≥2 insertions in the middle. The spread in fitness and read counts was very large, suggesting different causes for the existence of insertions in essential genes. (B) Transposon insertion locations, fitnesses, and projected ribosome binding site (RBS) strengths associated with intra-gene start codons. Black dots show all TA locations, and top row diamonds show the observed insertions with color coded fitness. The bottom diamonds represent alternative start codons in each gene. The color scales logarithmically with the associated RBS strengths. Each couple of rows represents the RBS strengths of all intra-gene start codons for both possible insertion orientations for each mutant, with the first row representing no insertion.
Table 1.
Comparison of gene essentiality calls between TIFA and DEC.
Figure 4.
Insertion frequency within genes.
Genes were evaluated in 0.2% gene increments. No insertion preference was observed in the complete gene population (A). The same analysis was performed (1% increments) after essential (red) and nonessential (green) gene identification (B). The last ∼2% at the 3′ end of essential genes were inserted more frequently.
Table 2.
Comparison of TIFA gene essentiality and FBA predictions.
Figure 5.
Comparison of gene essentiality between TIFA calls and FBA predictions.
Genes were grouped in 40 bins based on their cumulative probability values. The blue bars represent the genes not present in the MR-1 model. The red and green bars represent the FBA essential and FBA nonessential genes. Genes with a lower cumulative probability value are more likely to be essential, and genes with a higher cumulative probability value are more likely to be nonessential. The leftmost bar in (A) shows the number of genes with a cumulative probability value of less than the cut-off value (TIFA essential, allowing for 1 false positive). The rightmost bar in (B) shows the number of genes with a cumulative probability of more than the nonessential cut-off value (TIFA nonessential, allowing for 1 false positive).
Figure 6.
Three-way comparison between FBA predictions and TIFA and DEC essentiality calls.
(A) Comparison between FBA and TIFA. (B) Comparison between FBA and direct gene essentiality calls (DEC). The sectors of the comparison matrix show the intersections of FBA essential genes (E), FBA nonessential genes (N) and genes not in the MR-1 model (NA) with TIFA and DEC calls for essential genes (E), nonessential genes (N) and uncalled genes (U). The TIFA essential gene calls that could be alternatively explained by polar operon effects are shown in brackets.