* <2023-03-07 Tue> Tricking
Bkick 5
Aerial 5
front kick 7
aerial + front kick 5
scoot + bkick 5
tornado 5
scoot + aerial 5
feilong 6

On examine avec Julia

******** DONE Répartition par chromosome: variable mais seul le 6 = hotspot
CLOSED: [2023-03-07 Tue 22:44]

#+begin_src julia
using CSV, DataFrames
function cleanCount(fname)
    f = DataFrame(CSV.File(fname, header=false))
    transform!(f, :Column1 => ByRow(x -> string.(split(x, ":"))[1]) => :chrom)
    transform!(f, :Column1 => ByRow(x -> string.(split(x, "-"))[2]) => :pos)
    select!(f, Not([:Column1]))
    rename!(f, :Column2 => :nb)
end

Est-on dans des exons ?

f = cleanCount("count.csv")
grouped = groupby(f, :chrom)
for (k, g) in pairs(grouped)
    println("$(k.chrom): $(size(g, 1))")
end
#+end_src
#+RESULTS:
: NC_000001.11: 448
: NC_000002.12: 307
: NC_000003.12: 228
: NC_000004.12: 175
: NC_000005.10: 237
: NC_000006.12: 1223
: NC_000007.14: 274
: NC_000008.11: 414
: NC_000009.12: 161
: NC_000010.11: 194
: NC_000011.10: 304
: NC_000012.12: 254
: NC_000013.11: 168
: NC_000014.9: 184
: NC_000015.10: 250
: NC_000016.10: 239
: NC_000017.11: 373
: NC_000018.10: 85
: NC_000019.10: 576
: NC_000020.11: 83
: NC_000021.9: 67
: NC_000022.11: 117
******** Est-on dans des exons ?

#+begin_src sh
wget https://ftp.ncbi.nlm.nih.gov/refseq/H_sapiens/annotation/GRCh38_latest/refseq_identifiers/GRCh38_latest_genomic.gff.gz
awk '/BestRefSeq\texon/ && /transcript_id=NM/ {print $1"\t"$4"\t"$5;}' GRCh38_latest_genomic.gff | grep ^NC | save exons.csv
#+end_src
#+begin_src julia
using CSV, DataFrames
function groupedCount(fname)
    f = DataFrame(CSV.File(fname, header=false))
    transform!(f, :Column1 => ByRow(x -> string.(split(x, ":"))[1]) => :chrom)
    transform!(f, :Column1 => ByRow(x -> string.(split(x, "-"))[2]) => :pos)
    select!(f, Not([:Column1]))
    rename!(f, :Column2 => :nb)
    groupby(f, :chrom)
end

function groupedExons(fname)
    f = DataFrame(CSV.File(fname, header=["chrom", "start", "end"], delim="\t"))
    transform!(f, [:start, :end] => ByRow((x,y) -> Interval(x, y)) => :interval)
    select!(f, Not([:start, :end]))
    groupby(f, :chrom)
end

f = groupedCount("count.csv")
exons = groupedExons("exons.csv")
g = 1
x = parse(Int64, f[g][1, :pos])
x in exons[g][1, :interval]
#+end_src