170 haplo group assignment

malariagen / ag1000g-phase1-vgsc-report

MIT License

1 stars 7 forks source link

170 haplo group assignment #172

Closed cclarkson closed 6 years ago

cclarkson commented 6 years ago

Hi @alimanfoo

Fixed the F3/4 cluster mix-up and cleaned up the repository.

all analysis notebooks now run from combined haps_phase1.npz
all clustering membership estimated within repo
data/ directory cleaned

Standardising the clustering used for all analyses has altered some results quantitatively, could you please eyeball figures for haplotype homozygosity and assay design just to make sure nothing odd has happened that I have missed.

If this is all good and gets merged, I will make sure the paper builds okay (that I haven't changed any figure file names etc) and that all the text fits with the clustering we have used.

Resolves #170.

alimanfoo commented 6 years ago

Epic. Will take a look asap.

On Wed, 1 Aug 2018, 16:56 Chris Clarkson, notifications@github.com wrote:

@cclarkson https://github.com/cclarkson requested your review on: malariagen/agam-vgsc-report#172 https://github.com/malariagen/agam-vgsc-report/pull/172 170 haplo group assignment.

— You are receiving this because your review was requested. Reply to this email directly, view it on GitHub https://github.com/malariagen/agam-vgsc-report/pull/172#event-1765599499, or mute the thread https://github.com/notifications/unsubscribe-auth/AAq8QgPjX2DhZQT8eyOf8suaIBz_pxdVks5uMc-8gaJpZM4VqxPO .

cclarkson commented 6 years ago

Hi @alimanfoo

I rebuilt ms and everything looks good.

Just two comments:

should we add a paragraph on hierarchical clustering to the methods.
the hierarchical vs. median joining concordance had changed, but when I looked back I couldn't see where we had got 97% from. In the network notebook in master it was 95.4% and when I re-run it is 95.03%. Could you just check I am right please!