mschatz
commented
11 months ago It is very, very surprising to me that there are no kmers with coverage above 255. With a mode coverage of about 50x, this means that there are no repeats that occur more than 5 times in the genome which is shocking for an insect. Can you try running kmc on the raw HiFI data? Perhaps something went off with the kraken filtering
Good luck
Mike
On Tue, Jul 25, 2023 at 9:21 PM amv33576 @.***> wrote:
Hello,
I have HiFi data for a diploid insect, with an estimated genome size of ~425 Mb according to flow cytometry data. I used Kraken to filter reads and make sure there were no adapters in the reads. Then, I used kmc for the counts and get the frequency data:
kmc -k21 -t20 -m64 out.fastq 21mer ./kmc_tmp/ for the counts kmc_tools transform 21mer histogram 21mer.histo -cx255
I used cx 255, because I noticed after that number, everything was zero.
Finally, I run in the conda environment: genomescope2 -i 21mer.histo -o gs.kraken.def -k 21 --testing -m 1000. That resulted in panel A below. The predicted genome size seems close to what I expected, but genomescope calculated kcov=34.3 leaves me with three peaks. [image: genoscope2_output] https://user-images.githubusercontent.com/7957695/256071045-1d738edb-a75e-44dd-aaf8-aed5f6b1f49c.png
If I force the program to ignore the first peak and set -l 50, then I get ~ half of the genome size, as seen in Panel B. Also, the yellow line predicting the "unique sequence" are striking different between both panels as well.
So, what could explain these distributions, especially in panel A, where three peaks are observed? Is it peak 1-2 (34.3 and 68.6), one very broad peak? If yes, what could explain, or are 1-2 the Heterozygous/Homozygous peaks, then what is peak 3 representing?
I appreciate leads to understand these plots and get a better idea of the nature of the data.
— Reply to this email directly, view it on GitHub https://github.com/schatzlab/genomescope/issues/103, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABP34YFVLE37JKIB2XE3S3XSBWI5ANCNFSM6AAAAAA2X3E5VY . You are receiving this because you are subscribed to this thread.Message ID: @.***>
amv33576
tbenavi1
Hello,
I have HiFi data for a diploid insect, with an estimated genome size of ~425 Mb according to flow cytometry data. I used Kraken to filter reads and make sure there were no adapters in the reads. Then, I used kmc for the counts and get the frequency data:
kmc -k21 -t20 -m64 out.fastq 21mer ./kmc_tmp/ for the counts kmc_tools transform 21mer histogram 21mer.histo -cx255
I used cx 255, because I noticed after that number, everything was zero.
Finally, I run in the conda environment: genomescope2 -i 21mer.histo -o gs.kraken.def -k 21 --testing -m 1000. That resulted in panel A below. The predicted genome size seems close to what I expected, but genomescope calculated kcov=34.3 leaves me with three peaks.
If I force the program to ignore the first peak and set -l 50, then I get ~ half of the genome size, as seen in Panel B. Also, the yellow line predicting the "unique sequence" are striking different between both panels as well.
So, what could explain these distributions, especially in panel A, where three peaks are observed? Is it peak 1-2 (34.3 and 68.6), one very broad peak? If yes, what could explain, or are 1-2 the Heterozygous/Homozygous peaks, then what is peak 3 representing?
I appreciate leads to understand these plots and get a better idea of the nature of the data.