deploid-dev/mixedibd

mixedibd's Issues

need to rethink this

However, for samples with closely related strains, particularly at low coverage, only \texttt{DEploidIBD} can identify the correct number of strains and their proportions.

check pairwise IBD track length

change relatedness to rho, pearson correlation to r

make figures

• 1 alt vs ref IBD hap
• 2 validation plots
• 3 alt vs ref IBD + IBD length
• 4 vs MAP
• 5 Modelling
• 6 Haplotype painting

funding info from all partners

figure 3, add size for dots to show sample size

get files off, before rescomp shut down

create corrected IBD probs file to adjust for effective K

update bib

validation update

remove everything weighted analysis, both correlation and simple linear regression

Update figure 3 with filtering out bad samples

Fig3.pdf

Correlation between mixed IBD mean before and after filtering out bad samples

t = 10.032, df = 12, p-value = 3.46e-07
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
0.8318345 0.9828778
sample estimates:
cor
0.9452318

update figure 3 legend with standard country code

redo table

columns:
country, year, pfpr_include_zero, pfpr_exclude_zero, site, sample size, mean read, effective k, relatedness

DEploid-Lasso section in supplementary material

Action list

JAG

• Contact Ricard for pf6 meta data
• Filter low-quality haplotypes

JZ

• filterout mixed species samples
• check pf crosses envolope
• Repeat mixed of 2 validation using Africa samples, use 44/55, 25/75, 10/90 mixings with 3 levels of relatedness
• Figure 1
• Figure 2
• Figure 3
• Figure 5

JH

• IBD vs. MI IBD violin plot
• Classify all mixed samples from bg IBD

JZ + JH

• Figure 4

change figure 2

figure 2 and figure 2 supplement

figure2

• fig
• caption

figure 4

• fig
• caption

figure 5

• fig
• caption

ibd painting redo

fig3

• fig
  • no more nigeria
  • change shade from black to white
  • update collapsing haplotypes
• caption

figure1

• fig
• caption

vivax deconvolution

recompute relatedness

adding explanations for -ibd -lasso and -best flags

include mixed of threes figures

pfpr update

supplement, include figure to explain error

elife structure

it goes as

Abstract
eLife digest (https://en.wikipedia.org/wiki/ELife#eLife_digests)
Introduction
Results
Discussion
Materials and methods

Decision letter
Author response

redo figures

• make two copies Figures3
• Fig3
• Fig3sup
• Fig4
• Another copy of Fig4
• Fig5
• Another copy of Fig5

Before and after filtering

Figure 3 before filtering

Figure 3 after filtering

figure 3, move nigera to one side

change figures from abc to ABC

write supplement

check africa simulated haplotypes

for the panel,

> mean(aa)
[1] 1060.832
> sd(aa)
[1] 85.88517

check table 1 numbers

PF0180-C strange sample, chromsome 14 only inferred 400 alt calls (1000 on average from the same population)

From the Pf3k meta data, it seems pretty normal,

      sample       acc study
454 PF0180-C ERS017388  1017
                                                     study_title
454 Population genetics of natural populations in Northern Ghana
          contact_name        contact_email country     site collection_year
454 Lucas Amenga-Etego [email protected]   Ghana Navrongo            2010
    sample_prep      bases bases_mapped bases_duplicated avg_read_length
454             4724041788   4631818296        855311135              75
    mean_base_quality bases_of_1X_coverage bases_of_5X_coverage
454              25.5             23156910             23082804
    bases_of_10X_coverage bases_of_50X_coverage mean_coverage
454              23005542              21841663        196.64
    mean_fragment_size sd_fragment_size X.callable IsFieldSample
454              323.1             45.2       89.9          True
    PreferredSample AllSamplesThisIndividual
454            True                 PF0180-C

DEploid result seems normal

           ID Population    Site  Depth Est_K P1 P2 P3 P4   group relatedness
1510 PF0180-C      Ghana Kassena 196.64     1  1  0  0  0 Africa2          NA
     N50_IBD_Length.kb Mean_IBD_goodness cluster year eff_k
1510                NA                NA    <NA> 2010     1

we should acknowledge vikki