For future users who end up here, run this script:

bash format_GCT.sh norm_counts.txt output_prefix

Script assumed you have saved normalised counts from DESeq2 in R:

norm_counts <- counts(dds, normalized=TRUE)
write.table(norm_counts, file="norm_counts.txt", append = TRUE, sep="\t", row.names = TRUE, col.names = TRUE, quote = FALSE)

Here is the format_GCT.sh bash script:

#!/usr/bin/env bash

## Script to convert DESeq2 normalized counts to GCT format

if [ "$1" == "-h" ]; then
  echo "Usage: bash `basename $0` [FILE] [OUT_PREFIX]"
  echo
  echo "Example: bash `basename $0` norm_counts.txt GSE118959"
  exit 0
fi

if [ "$1" != "" ]; then
    echo "Counts File: $1"
else
    echo "Counts File is missing"
fi

if [ "$2" != "" ]; then
    echo "Output Prefix: $2"
else
    echo "Output Prefix is missing"
fi

FILE=$1
OUT=$2

TMP=`wc -l $FILE | awk '{print $1}'`
LEN=`expr $TMP - 1`
TMP=`awk '{print NF}' $FILE | sort -nu | tail -n 1`
SAMPLES=`expr $TMP - 1`


cat $FILE \
| awk -F"\t" '{if(NR==1) $1="NAME"FS$1}1' OFS="\t" \
| awk '{$1 = $1 OFS (NR==1?"Description":"na")}1' \
| sed 's/ /\t/g' \
| awk -v LEN=$LEN -v SAMPLES=$SAMPLES \
 'BEGIN{print "#1.2""\n"LEN"\t"SAMPLES}1' > ${OUT}.gct

You are skipping an awful lot of steps of your analysis. Your fastq data first needs to be aligned to the genome, reads need to be counted and you need to perform differential expression analysis. The results of the last step can be used in GSEA.