songqiang

from pandas import crosstab
def pyfreq(v1, v2 = None, weight = None, order = ''):
    if v2 is not None:
        if weight is not None:
            cnt = crosstab(v1, v2, weight, aggfunc = sum, margins = True, dropna = False)
        else:
            cnt = crosstab(v1, v2, margins = True, dropna = False)

        if order == 'freq':
            cnt.iloc[:-1, :] =  cnt.iloc[:-1, :].sort_values('All', ascending = False).values

songqiang

fromFile=Mouse_ESC.meth
indexFile=CpGs-mm9Tohg19.bed
toFile=Mouse_ESC-hg19.meth

$fastLiftOver -f $fromFile -i $indexFile -t $toFile -v

# sort output and merge sites mapped to the same location
export LC_ALL=C;
tmpfile=$(mktemp);
wc -l $toFile

songqiang

promoters=~qiangson/panfs/data/human/hg18/gene/promoter-nr.bed

CpGs=~qiangson/panfs/co-methylation/methylomes/Berman-Human-2012-Human_ColonCancer-results-Human_ColonCancer_Meth.bed

~qiangson/app/bedtools/bin/intersect $promoters $CpGs \
    |cut -f4-6|uniq -c \
    |awk '{print $2,$3,$4,2*$1/($4-$3);}' > CpG-densities.bed