De Bruijn graphs are structures that appear naturally in the study of strings. Therefore the rise of de Bruijn graph based sequence analysis approaches is not a surprise. The problem with de Bruijn graphs is that for most of their applications in Bioinformatics they are too large even for small genomes. A way to overcome this problem is the compression of branch-free paths to single nodes. Although this compression is a common first step in many of the de Bruijn graph based approaches, its direct construction from raw data does not seem to be documented before. Our experience shows that, though based on simple operations, implementing the construction of such graphs is a tricky and time consuming task. Therefore we shortly describe in this report our graph construction algorithm and hope that the given details are enough to help the reader skipping some pitfalls we found by doing this task.