Jtdylktuy

I have a table data like below

abc 1   1   1

bcd 2   2   4

bcd 12  23  3

cde 3   5   5

cde 3   4   5

cde 14  2   25

I want the sum of values in each column based on variables in first column and desired result is like below:

abc 1   1   1

bcd 14  25  7

cde 20  11  35

I used awk command like this

awk -F"t" '{for(n=2;n<=NF; ++n)a[$1]+=$n}END{for(i in a ) print i, a[i] }' tablefilepath

and I got a result below:

abc 3

bcd 46

cde 66

I think the end of my code is wrong but don't know how to fix it.
I need some directions to fix the code.

You were fairly close. 
You see what you were doing wrong, don't you? 
You were keeping one total for each column 1 value,
when you should have been keeping three.

This is similar to Inian's answer,
but trivially extendable to handle any number of columns:

awk -F"t" '{for(n=2;n<=NF; ++n) a[$1][n]+=$n}

        END {for(i in a) {

                printf "%s", i

                for (n=2; n<=4; ++n) printf "t%s", a[i][n]

                printf "n"

             }

        }'

Rather than keep three arrays, like Inian's answer,
it keeps a two-dimensional array.

So long as your file is tab-delimited, datamash is a good fit for this.

$ datamash groupby 1 sum 2 sum 3 sum 4 < tablefilepath

abc     1       1       1

bcd     14      25      7

cde     20      11      35

Datamash can also work with non-tabs, if you specify -t <delimiter>. But tabs seem closest to the example input you have provided.

Datamash won't work if your input is delimited by arbitrary whitespace (i.e. possible multiple spaces intended to "look like" a tab). Still, even if that is what your data looks like, it is easily munged into the form expected by datamash:

sed -i 's/ +/t/g' tablefilepath

Using awk summing up the columns 2-4 based on 1.

awk -v FS="t" -v OFS="t" '{ col1[$1]+=$2; col2[$1]+=$3; col3[$1]+=$4; next } END { for ( i in col1) print i, col1[i], col2[i], col3[i]  }' file