# Aggregate the CKP25 mouse data:
library(cbrbase)

files = c('200812TsaA_aggregated.counts.tsv',
        '200909Tsa_aggregated.counts.tsv',
        '200916Tsa_aggregated.counts.tsv',
        '200923Tsa_aggregated.counts.tsv',
        '200930Tsa_aggregated.counts.tsv',
        '201007Tsa_aggregated.counts.tsv',
        '201021Tsa_aggregated.counts.tsv',
        '201014Tsa_aggregated.counts.tsv')

adf = NULL
for (fn in files){
    df = read.delim(fn, sep="\t")
    print(fn)
    print(dim(df))
    if (is.null(adf)){
        adf = df
    } else { 
        adf = cbind(adf, df[,-1, drop=F]) 
    }
}

rn = adf$gene
gind = grep("^ENSM",rn)
amat = as.matrix(adf[gind,-1])
cn = sub("\\.","-",sub("^X","",colnames(amat)))

write.table(adf, 'all_aggregated.counts.tsv',quote=F, sep="\t", row.names=F)
write.table(amat, 'all_aggregated.counts_mat.tsv',quote=F, sep="\t", row.names=F, col.names=F)
write.table(rn[gind], 'all_aggregated.rn.tsv',quote=F, sep="\t", row.names=F, col.names=F)
write.table(cn, 'all_aggregated.cn.tsv',quote=F, sep="\t", row.names=F, col.names=F)

tab = read.delim('../Annotation/gencode.vM25.basic.annotation.genes.tsv', header=F)[,c(9,11)]
names(tab) = c('gene','symbol')
rownames(tab) = tab$gene
sn = tab[rn[gind],'symbol']

write.table(sn, 'all_aggregated.rn.symb.tsv',quote=F, sep="\t", row.names=F, col.names=F)