Steps.txt

Steps in the tsinfer analysis

1) Merging vcf files by samples
    • Start with .vcf.gz
    • Extract sampleIDs for each file for each chromosome
    • Compare files and secure there are no overlaps
    • Filter vcfs to remove overlapping sampleIDs 
(I guess this could become an if statement)
    • Compress and index
    • Merge files
    • what are the checks here? The main point here is not having duplicated samples because they will become a problem further down the line
        ◦ if we get an error with chromosome number > 23, then rename to 1 and rename back after the analysis

2) Prepare Files for tsinfer
    • we start with compressed and indexed vcf files
    • whether we start with one or multiple vcfs –if there is only vcf → we split it into chromosome vcfs
    • add INFO (allele frequency, allele count) – extract ALT, REF, AF → major allele file → put it in the ancestral alleles (this needs to be done after merging)we decompress the files
    • next, we put in the ancestral information
        ◦ the ancestral file need to be standardized
        ◦ for the sites that don’t have an ancestral allele known, we leave a black space in the VCF (make sure this script runs fine and do what expected!)
    • - we compress and index the files

3) Phasing
    • If files are compressed and indexed, it is just calling SHAPEIT4
    • genetic map (check whether the maps are in the correct and same format)
    • what are the checks here? Ensure there is only one file/chromosome 
Ensure that files are per chromosome, check the output, make sure there is enough memory

4) Infer the trees
    • prepare the samples file
    • what steps do each of us take: have the three steps for all the inferences
    • decide how a standard metafile should look like
    • infer by chromosome (this is probably true for everyone)
    • what about tsdate?? Are we putting here as well or having a separate one? 
One issue I constantly have is (if sample to dated.tree are all on the same scrips) is when it crash because of wrong time/memory set ups and I have to run the full script! But I guess that is the point of snakemake so we are probably good having it all together?? 

5) Do statistics on the trees
    • GNN
    • Fst
    • Tajima’s D
    • IBD
    • GRM ...

6) Tsdate
7) Ne