mnshgl0110
commented
2 years ago I am curious about the initial flipping and errors. I can imagine when the warning is useless, but did you find chromosomes that required flipping but SyRI did not give any warning for them?
Nevertheless, this is a very nice example to illustrate a current limitation of SyRI.
SyRI expects that the homologous chromosomes being compared would be closely related i.e. largely syntenic. It performs an initial check to see if number of inversely aligned bases is more than directly aligned bases for a reference chromosome. If yes, then it warns to ensure same strandedness, but this warning does not mean that the chromosome must be reverse complemented as currently SyRI cannot identify that.
Too-many inverted alignments could mean that different strands are being compared resulting in no identified synteny and subsequent crashes. However, it is possible (as here) that homologous chromosomes with same strands have many inversions, in which case the chromosomes should not be flipped, but SyRI could still give the warning.
It would be great to have an automatic identification of which chromosomes to flip and I intend to that, but I am lacking time to work on it. If you can set it up and would like to share the code, then I can add that to SyRI so that it does that directly from the BAM files itself.
kaede0e
tiramisutes
aaannaw
Hi, We have been using SyRI for two-genome comparison in multiple species pairs, and issue #48 has been helpful in fixing problems whenever "Index out of range" error shows up. The chromosomes to reverse complement (flip) usually gets identified by SyRI as a warning that looks something like this: "Reading Coords - WARNING - Reference chromosome CM018900.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014298.1). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors." So in this case, I would flip CM014298.1 and try again, and for the most part this has been working.
However, we encountered a weird situation with this particular pair we were working on(table_original.txt.gz - this is coords file for the minimap2 alignment with the unreversed genome). The usual troubleshooting step above did not work and SyRI couldn't run with the flipped genome. We were curious about what went wrong and so visualized this coords file (coords_file_for_visualization_original.v0.pdf), and found out that chromosomes SyRI told us to flip were incorrect.
We then manually selected chromosomes to flip based on the visualized plot, which was easy to identify, and re-ran SyRI. This time it worked. But weirdly enough, this manual flipping still produced the warning:
Reading Coords - WARNING - Chromosomes IDs do not match. Reading Coords - WARNING - Matching them automatically. For each reference genome, most similar query genome will be selected. Check mapids.txt for mapping used. Reading Coords - WARNING - Reference chromosome CM018888.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014286.1/rc). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018891.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014289.1). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018893.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014291.1/rc). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018894.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014292.1/rc). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018895.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014293.1/rc). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018896.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014294.1/rc). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018897.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014295.1/rc). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018900.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014298.1). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018901.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014299.1). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018903.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014302.1). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018905.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014304.1/rc). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018908.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014306.1). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors. Reading Coords - WARNING - Reference chromosome CM018910.1 has high fraction of inverted alignments with its homologous chromosome in the query genome (CM014308.1/rc). Ensure that same chromosome-strands are being compared in the two genomes, as different strand can result in unexpected errors.
This points out that some chromosomes SyRI automatically recognizes and suggests to reverse complement are not correct. We are wondering why this happens, and if it is something that comes up frequently. How does SyRI recognize the "high fraction of inverted alignments"? We are currently trying to automate the whole process, and seeking for a way around it so that we don't have to manually select chromosomes to flip from the visualization tools.
Thank you for your feedback, Kaede