RFMix_PopPhased skipping growing forests

[angelcymak]

I am running RFMix version 1.5.4. When I run PopPhased, I would normally see these messages shown on the screen:

Giddyup Reading files Number of processors available: 24 Minimum node size: 1 Number of SNPs Read: 22676 Number of SNPs to Exclude: 0 Number left after pruning: 22676 Number of windows: 800 Creating phasings Done creating phasings Growing forests Applying Conditional Random Fields CRF on individual: 0 CRF on individual: 1 CRF on individual: 2 CRF on individual: 3 CRF on individual: 4 .... CRF on individual: 2180 Checking input... Done checking input

However, I am having problem with one of the chromosome that the steps from "Growing forests" and "Applying Conditional Random Fields" are skipped without any errors. Could you please advise what may be wrong?

[slowkoni]

Hi Angel,

I’m sorry, I’m unable to answer your questions below. I wrote and support RFMIX v2.X available on GitHub under github.com/slowkoni/rfmix http://github.com/slowkoni/rfmix. RFMIX 2.X more or less implements the same methodology originally developed by Brian Maples at Stanford during his graduate research. I believe he maintains a website which makes his original final version, 1.5.4 that you refer to below, available somewhere, and obviously, you have found that, but you would have to contact him for help. Unfortunately, I do not have a current email address for him. I think the last few times I tried his old Stanford address (he graduated some years ago) the email just bounced. My stanford email address will also bounce as I no longer work there.

Please only open GitHub issues/tickets for RFMIX 2.X if they pertain to that version series and source code base, which is completely different than the one Brian originally developed. For 1.5.4, I suggest trying to find Brian on LinkedIn perhaps, he’s out there somewhere. Or, you can try RFMIX 2.X if you are an academic or research institute user. For commercial use including non-profit organizations a license is required, please contact Stanford Office of Technology Licensing for that.

Best, Koni

On Jul 20, 2018, at 15:47, Angel Mak notifications@github.com wrote:

I am running RFMix version 1.5.4. When I run PopPhased, I would normally see these messages shown on the screen:

Giddyup Reading files Number of processors available: 24 Minimum node size: 1 Number of SNPs Read: 22676 Number of SNPs to Exclude: 0 Number left after pruning: 22676 Number of windows: 800 Creating phasings Done creating phasings Growing forests Applying Conditional Random Fields CRF on individual: 0 CRF on individual: 1 CRF on individual: 2 CRF on individual: 3 CRF on individual: 4 .... CRF on individual: 2180 Checking input... Done checking input

However, I am having problem with one of the chromosome that the steps from "Growing forests" and "Applying Conditional Random Fields" are skipped without any errors. Could you please advise what may be wrong?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/slowkoni/rfmix/issues/6, or mute the thread https://github.com/notifications/unsubscribe-auth/AJapKdJsgUkN2OsXUssFtD6WCCSLW8jdks5uIl4dgaJpZM4VZXHD.

[angelcymak]

Hi Koni,

Thanks very much for your response. I will try the latest RFMix version and bug you more if I have other questions.

Thank you, Angel

On Sat, Jul 21, 2018 at 4:04 PM, Mark Koni Wright notifications@github.com wrote:

Hi Angel,

I’m sorry, I’m unable to answer your questions below. I wrote and support RFMIX v2.X available on GitHub under github.com/slowkoni/rfmix < http://github.com/slowkoni/rfmix>. RFMIX 2.X more or less implements the same methodology originally developed by Brian Maples at Stanford during his graduate research. I believe he maintains a website which makes his original final version, 1.5.4 that you refer to below, available somewhere, and obviously, you have found that, but you would have to contact him for help. Unfortunately, I do not have a current email address for him. I think the last few times I tried his old Stanford address (he graduated some years ago) the email just bounced. My stanford email address will also bounce as I no longer work there.

Please only open GitHub issues/tickets for RFMIX 2.X if they pertain to that version series and source code base, which is completely different than the one Brian originally developed. For 1.5.4, I suggest trying to find Brian on LinkedIn perhaps, he’s out there somewhere. Or, you can try RFMIX 2.X if you are an academic or research institute user. For commercial use including non-profit organizations a license is required, please contact Stanford Office of Technology Licensing for that.

Best, Koni

On Jul 20, 2018, at 15:47, Angel Mak notifications@github.com wrote:

I am running RFMix version 1.5.4. When I run PopPhased, I would normally see these messages shown on the screen:

Giddyup Reading files Number of processors available: 24 Minimum node size: 1 Number of SNPs Read: 22676 Number of SNPs to Exclude: 0 Number left after pruning: 22676 Number of windows: 800 Creating phasings Done creating phasings Growing forests Applying Conditional Random Fields CRF on individual: 0 CRF on individual: 1 CRF on individual: 2 CRF on individual: 3 CRF on individual: 4 .... CRF on individual: 2180 Checking input... Done checking input

However, I am having problem with one of the chromosome that the steps from "Growing forests" and "Applying Conditional Random Fields" are skipped without any errors. Could you please advise what may be wrong?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/slowkoni/rfmix/issues/6>, or mute the thread < https://github.com/notifications/unsubscribe-auth/ AJapKdJsgUkN2OsXUssFtD6WCCSLW8jdks5uIl4dgaJpZM4VZXHD>.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/slowkoni/rfmix/issues/6#issuecomment-406828919, or mute the thread https://github.com/notifications/unsubscribe-auth/AG7H8M2saRvKEsyqF_A1xMr6WzG6u3Qiks5uI7OFgaJpZM4VZXHD .

[dshriner]

Hi Angel,

I have encountered a similar problem. It appears to be an out-of-memory issue. One workaround is to split the sample into batches of individuals. Another option is to decrease the window size.

Thanks, Daniel

[angelcymak]

Got it! Thanks Daniel.