Evidence for joins in contigs

[donkirkby]

Sometimes, assembled contigs look strange, and you wonder whether they are real or a misassembly. Could we add a display to show evidence for some of the strange features in contigs? The first two features I can think of are large deletions and contig sections in a different order from the reference. I found examples of each in the 22 Oct 2021 run: Y04280A-HIVB_S10 has both. Contig 1 has a large deletion of most of gag that is covered by contig 2. Contig 1 also has a section out of order with 1.3 aligning in the 5' LTR. (I wonder if 1.3 should actually align in the 3' LTR, and then it would just be a large deletion of GP41 covered by contig 3.) Y04279A-HIVB_S9 has almost the same deletion in gag, but without the other problem.
In the remapped version, we get clues from the amount of soft clipping when the reads don't agree with the reference. Seeing that displayed across the whole contig might be helpful. I also wonder if it would be helpful to show the bias toward right or left at each point in the contig. If reads are random fragments of the genome, then we'd expect any point away from the ends to be covered by roughly the same number of reads extending farther left than right as those that extend farther right than left. As you come close to the ends of the genome or close to a misassembly, you'd expect more reads to extend into the correct assembly than would extend across the misassembly or off the end.
Maybe plotting the bias with a diverging colour map like bwr would be helpful. Blue could show left bias and red could show right bias. Alternatively, just use a line plot of right bias from 0.0 to 1.0 across each contig.