Completely different result between Genomescope v1 and v2 #48

New issue

Open

Completely different result between Genomescope v1 and v2#48

ptranvan

Hi,

I used genomescope from this repo:

https://github.com/schatzlab/genomescope
and
https://github.com/tbenavi1/genomescope2.0

And got weird results:

From v1:

https://ibb.co/FXPYPgV

From v2:

https://ibb.co/XC7Y5Nx

What could explain those big differences ?

hist have been generated with:

kmc -k21 -t5 -m64 -ci1 -cs10000 @cat.files cat.kmcdb tmp

kmc_tools transform cat.kmcdb histogram cat.kmcdb_k21.hist -cx10000

mschatz

Contributor

This is one of those challenging situations where it is ambiguous if it is a larger haploid genome with low heterozygosity (v1, 463Mb @ .113% heterozygosity) versus a smaller haploid genome with very high heterozygosity (v2, 251Mb @ 16.1% heterozygosity) Do you know anything about the genetic architecture of this? My blind guess is the v1 model is probably accurate since 16.1% heterozygosity would be extremely high from a single individual. You can help inform the genomescope 2.0 model to give a very similar result by giving it a hint that the "Average k-mer coverage for polyploid genome" to be 114. Good luck! Mike

…

On Fri, Dec 11, 2020 at 8:25 AM Patrick Tran Van ***@***.***> wrote: Hi, I used genomescope from this repo: https://github.com/schatzlab/genomescope and https://github.com/tbenavi1/genomescope2.0 And got weird results: From v1: https://ibb.co/FXPYPgV From v2: https://ibb.co/XC7Y5Nx What could explain those big differences ? hist have been generated with: kmc -k21 -t5 -m64 -ci1 -cs10000 @cat.files cat.kmcdb tmp kmc_tools transform cat.kmcdb histogram cat.kmcdb_k21.hist -cx10000 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#48>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABP345FHTXZPXVLXJS6VDDSUIMTNANCNFSM4UWUFEDA> .

ptranvan

Author

Hello, we don't know many things about the genomic architecture of this species but it should be diploid. I did change "Average k-mer coverage "for polyploid genome" to 114 but got the same plot:

GenomeScope version 2.0
input file = user_uploads/va4PdkFoOZtVXygmISbA
output directory = user_data/va4PdkFoOZtVXygmISbA
p = 2
k = 21
initial kmercov estimate = 114

property min max
Homozygous (aa) 0% 100%
Heterozygous (ab) 0% 100%
Genome Haploid Length 251,713,776 bp 251,965,450 bp
Genome Repeat Length 52,929,734 bp 52,982,656 bp
Genome Unique Length 198,784,041 bp 198,982,794 bp
Model Fit 79.394% 92.8582%
Read Error Rate 0.236534% 0.236534%

http://genomescope.org/genomescope2.0/analysis.php?code=va4PdkFoOZtVXygmISbA

ptranvan

mentioned this

on Dec 15, 2020

HiFi data supported ? #49

mschatz

Contributor

Sorry, I should have said run it setting the average coverage to by 57 so that it will realize the tall peak at 117x is the homozygous kmer peak: http://qb.cshl.edu/genomescope/genomescope2.0/analysis.php?code=W2E0vPLDTdDqDT038A18 Cheers Mike

…

On Mon, Dec 14, 2020 at 5:51 AM Patrick Tran Van ***@***.***> wrote: Hello, we don't know many things about the genomic architecture of this species. I did change "Average k-mer coverage "for polyploid genome" to 114 but got the same plot: GenomeScope version 2.0 input file = user_uploads/va4PdkFoOZtVXygmISbA output directory = user_data/va4PdkFoOZtVXygmISbA p = 2 k = 21 initial kmercov estimate = 114 property min max Homozygous (aa) 0% 100% Heterozygous (ab) 0% 100% Genome Haploid Length 251,713,776 bp 251,965,450 bp Genome Repeat Length 52,929,734 bp 52,982,656 bp Genome Unique Length 198,784,041 bp 198,982,794 bp Model Fit 79.394% 92.8582% Read Error Rate 0.236534% 0.236534% http://genomescope.org/genomescope2.0/analysis.php?code=va4PdkFoOZtVXygmISbA — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#48 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABP342WCW67HOVYRBNLF3TSUXU4VANCNFSM4UWUFEDA> .

ViriatoII

I'm also curious about this. Genomescope1 seems to have estimations in line with literature for my species while genomescope2 not (even when multiplying by 2 because of haploid vs diploid)

mschatz

Contributor

Thats the usual issue. If you post the links to the results I can give more specific comments. Cheers Mike

…

On Thu, Aug 19, 2021 at 7:31 AM ViriatoII ***@***.***> wrote: I'm also curious about this. Genomescope1 seems to have estimations in line with literature for my species while genomescope2 not (even when multiplying by 2 because of haploid vs diploid) — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#48 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABP347E7LZJGISOLTJDSETT5TTR7ANCNFSM4UWUFEDA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .

ViriatoII

Hi Mike,
That's very kind of you, thank you.
As an example, this D. erucoides is estimated to have ~500 Mbps haploid genome size, 1000 Mbps in diploid size ( Lysák et al.,2009)

Genomescope1 predicts 435 Mbps haploid length, just short of literature, as well as a reasonable 0.65% of heterozygosity:
http://qb.cshl.edu/genomescope/analysis.php?code=izm95ZeGs1WxSvj8bidT

The genomescope2 run (even using max 100 000 coverage) predicts 212Mbps of haploid genome length, and a surprising max estimated heterozygosity of 20%.
http://qb.cshl.edu/genomescope/genomescope2.0/analysis.php?code=cKhv8wWNKBHMEgf3ilHd

I am applying the same pipeline to 20 species, some of which are tetraploid. I'd wish I could just apply the same parameters to all, or at least only treat the tetraploids differently and give them to genomescope2.

Appreciate the help,
Ricardo

mschatz

Contributor

Hi Ricardo It looks like GenomeScope2 got a little confused on which peak was which - notice in GS1 the first peak (kcov) is at 19, but in GS2 the first peak is at 38. In these cases, you can give the algorithm a hint as to which peak should be the first peak. On the website it is listed as "Average k-mer coverage for polyploid genome". If I give it at hint of 19, the GS2 results are very similar to the GS1 results: http://qb.cshl.edu/genomescope/genomescope2.0/analysis.php?code=2PR3qCNoIXJQ22mXMC6d Looking at this I think GS2 gets confused by the excess of kmers that have a coverage around 10-20 that dont look like normal sequencing errors. I would guess your sample has a somewhat unusual structure, such as some contamination or maybe unbalanced representation of sex chromosomes. It also looks like the genome size estimate is smaller than expected because the kmer distribution is truncated at 10000x coverage. If you increase this to 100,000 or 1,000,000 it will capture more of the higher order repeats and increase the genome size. Good luck! Mike

…

On Fri, Aug 20, 2021 at 4:37 AM ViriatoII ***@***.***> wrote: Hi Mike, That's very kind of you, thank you. As an example, this D. erucoides is estimated to have ~500 Mbps *haploid* genome size, 1000 Mbps in diploid size ( Lysák et al.,2009) Genomescope1 predicts 435 Mbps *haploid* length, just short of literature, as well as a reasonable 0.65% of heterozygosity: http://qb.cshl.edu/genomescope/analysis.php?code=izm95ZeGs1WxSvj8bidT The genomescope2 run (even using max 100 000 coverage) predicts 212Mbps of haploid genome length, and a surprising max estimated heterozygosity of 20%. http://qb.cshl.edu/genomescope/genomescope2.0/analysis.php?code=cKhv8wWNKBHMEgf3ilHd Appreciate the help, Ricardo — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#48 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABP347CL5ZGAGZ5EEZLWUTT5YH5LANCNFSM4UWUFEDA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .

to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

Labels

No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Completely different result between Genomescope v1 and v2 #48

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Participants

Completely different result between Genomescope v1 and v2 #48

Description

Activity

mschatz commented on Dec 14, 2020

ptranvan commented on Dec 14, 2020

mschatz commented on Dec 16, 2020

ViriatoII commented on Aug 19, 2021

mschatz commented on Aug 20, 2021

ViriatoII commented on Aug 20, 2021

mschatz commented on Aug 20, 2021

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Participants

Issue actions