malariagen / ag1000g-phase2-vgsc-report Goto Github PK

If there are really no alleles in phase 2, this can be removed. If there are, why haven't they made it into the display table?

When rendering the networks we get a warning, I'm guessing for each node, so it kicks out a lot of text -

Warning: gvrender_set_style: unsupported style normal - ignoring

• this will necessitate setting up data access
• extract the mutations in vgsc
• we need to know if any phase 1 sites have been filtered out
• we need to know where we have multi-allelics that will need 'special' phasing

While generating supplementary data files for phase 2, I have noticed that the multi-allelic at 2431061 produces two different amino acids - this should have two separate lines on the hierarchical clustering plot - check other analyses too.

Calculating number of 'double hets' is not a problem, however estimating deviation from expected/null hypothesis here is tricky.

We need to decide on tree cutting and cluster naming - this should be in line with phase 2.

The current cut off (set to recapitulate Phase 1) highlights four non-995 variant carrying clusters - these look to have interesting properties, perhaps we should highlight them.

pdf lines 214-216 discuss the diversity of cluster - update this in line with the notebook.

This would also be a good time to sanity check the notebook.

Is it possible to create a figure like the PBO/pyrethroid map but use ace-1 SNPs and CNVs

• Ask Eric if we can get vgsc-like CNV data for Ace-1

• Research ace-1 marker SNPs

• Combine into figure / report stats

Some time ago there was talk of removing the L1/L2 results and discussions completely from the manuscript - I have produced versions of figures with and without L1/L2.

I think this is a problem with petl - "rjust(6) error - see notebook

what is says on the tin

UnsatisfiableError: The following specifications were found to be incompatible with a past
explicit spec that is not an explicit spec in this operation (libgcc-ng):

• gdal -> libgcc-ng[version='>=7.3.0']
• gdal -> libstdcxx-ng[version='>=7.3.0']
• gdal -> numpy[version='>=1.14.6,<2.0a0']

The following specifications were found to be incompatible with each other:

Are we happy with the three multi-allelics I phased onto the haplotypes with MVN? Are there other variants we'd like to see?

I'm unclear as to why we included/excluded certain variants from the missense variants list - needs clarification.

Always a fiddle, but now it seems to have stopped working.

When plotting the dendrogram there is a warning about a deprecated functions - tick10n and tick20n. I think these come from agam-report-base. Track down and fix.

Repeat phase 1 analysis with phase 2 data.

• Check, are all P2 references updated?

• Line 156: “… for haplotypes carrying the L995 allele, ….”. From the context I believe this should be “L995F”

• Page 14. Positive selection on non-synonymous alleles: using EHH analysis the authors found significant differences in haplotype homozygosity between different genetic backgrounds carrying resistance alleles, suggesting differences in the timing and/or strength of selection. As the Ag1000 phase 2 dataset encompasses mosquito collection spanning over a decade (2000 to 2012), could the authors comment on the potential impact of the temporal sampling scheme on the observed selection events?

Re-run phase 1 code with phase 2 data

We now have this lovely looking protein structure figure, I am contacting Andrias about methods so we can integrate. The question is, where do we think it would sit best?

Re-run EHH analysis using the phase 2 dataset.

Re-run the networks analysis from the Phase 1 paper.

As included in phase 1

Canuscript currently has biorxiv citation.

Remove L1 and L2 from the network clustering file, then build map from this?

Now we have re-run all vgsc analysis with the phase 2 data set, the manuscript needs updating.