Thale cress#

Arabidopsis thaliana

Input files#

To run this example, you will need to download the following three input files:

There is only one command you need to run LiftOn:

lifton -g TAIR10.gff -o lifton.gff3 -copies ASM2311539v1_genomic.fna TAIR10.fna

After successfully running LiftOn, you will get the following file and output directory:



Results#

Genome annotation evaluation#

Here are some visualization results comparing LiftOn annotation to (1) Liftoff and (2) miniprot annotation.

First, we calculate the protein sequence identity score for every protein-coding transcript (check Evaluation metrics - sequence identity section) for three annotations, LiftOn, Liftoff, and miniprot.

Figure 23 compares the protein-coding gene mapping of Liftoff, based on DNA alignment, with miniprot, utilizing protein-to-DNA alignment. Dots in the lower right signify transcripts where Liftoff outperformed miniprot in protein sequence identity, while the upper left indicates transcripts where miniprot excelled. LiftOn employs the PM algorithm to enhance annotations in both, achieving improved protein-coding gene annotation, as neither approach dominates the other.

../../_images/parasail_identities4.png

Figure 23 The scatter plot of protein sequence identity comparing between miniprot (y-axis) and Liftoff (x-axis). Each dot represents a protein-coding transcript.#


Next, we individually assess LiftOn in comparison to Liftoff and miniprot. In the comparison of LiftOn versus Liftoff (Figure 24, left), 186 transcripts demonstrate higher protein sequence identity, with 35 achieving 100% identity. Similarly, in the LiftOn versus miniprot comparison (Figure 24, right), 10627 protein-coding transcripts exhibit superior matches, elevating 10466 to identical status relative to the reference.

../../_images/combined_scatter_plots4.png

Figure 24 The scatter plot of protein sequence identity comparing between LiftOn (y-axis) and Liftoff (x-axis) (left) and comparing between LiftOn (y-axis) and miniprot (x-axis) (right).#


We visualize the transcripts in a 3-D plot, incorporating LiftOn, Liftoff, and miniprot scores (see Figure Figure 25) to provide a comprehensive comparison of the three tools. If a dot is above the \(x=y\) plane, it indicates that the protein-coding transcript annotation of LiftOn generates a longer valid protein sequence aligning to the full-length reference protein. The 3-D plot reveals that the majority of dots are above the \(x=y\) plane, suggesting that LiftOn annotation is better.

../../_images/3d_scatter4.png

Figure 25 The 3-D scatter plot of protein sequence identity comparing between LiftOn (y-axis), Liftoff (x-axis), and miniprot (z-axis).#


Next, we check the distribution of protein sequence identities (see Figure 26). Among the three tools, LiftOn (middle) exhibits the smallest left tail, with 26 protein-coding transcripts having a protein sequence identity of \(< 0.4\).

../../_images/combined_frequency_log4.png

Figure 26 Frequency plots in logarithmic scale of protein sequence identity for Liftoff (left), LiftOn (middle), and miniprot (right) for the results of arabadop lift-over.#


Finding extra copies of lift-over features#

LiftOn also has a module to find extra copies by using intervaltree, Liftoff, and miniprot. The Circos plot in Figure 27 shows their relative positions between the two genomes. The plot illustrates that the extra copies were predominantly located on the same chromosomes in both TAIR10 and ASM2311539. The frequency plot of extra copy features are show in Figure 28.

../../_images/circos_plot4.png

Figure 27 Circos plot illustrating the locations of extra gene copies found on ASM2311539 (left side) compared to TAIR10 (right side). Each line shows the location of an extra copy, and lines are color-coded by the chromosome of the original copy.#


../../_images/frequency4.png

Figure 28 Frequency plot for additional gene copy.#


Finally, we examined the order of protein-coding genes (Figure 29) between the two genomes and observed that, as expected, nearly all genes occur in the same order and orientation in both human genomes.

../../_images/gene_order_plot4.png

Figure 29 Protein-gene order plot, with the x-axis representing the reference genome (TAIR10) and the y-axis representing the target genome (ASM2311539). The protein sequence identities are color-coded on a logarithmic scale, ranging from green to red. Green represents a sequence identity score of 1, while red corresponds to a sequence identity score of 0.#


What's next?#

Congratulations! You have finished this tutorial.

See also