Milea Timbergen

63 < 0.05, confidence interval 95%) between DTF with various CTNNB1 mutation types (T41A, S45F) and DTF with no mutations in exon 3 of CTNNB1 (WT). Selection of Wnt target genes To rule out that the limited set of Wnt genes analysed so far caused a selection bias affecting the outcome of our analyses, the number of Wnt targets genes was increased. Wnt target genes used in a hierarchical unsupervised cluster analyses using, transcript expression data from DTF tumours, were selected using the mammalian Wnt target genes listed on the Wnt homepage (https://web.stanford.edu/group/nusselab/cgi-bin/wnt/target_genes ) as a reference 35 . Supplemental Table 1 summarizes the mammalian Wnt target genes that were selected, the corresponding Affymetrix probe sets and alias terms (https://www.ncbi.nlm . nih.gov/gene/). Since DTF tumours are of mesenchymal origin, they may express Wnt target genes partly different from those observed in epithelial cancers. Therefore, unsupervised hierarchical cluster analyses were repeated using a list of putative mesenchymal Wnt target genes in desmoids published by Denys et al 36 . Gene names, corresponding Affymetrix probe sets and alias terms are shown in Supplemental Table 2. Hierarchical cluster analysis For hierarchical cluster analysis, the retrieved normalized mRNA expression data of the 128 DTF samples of the French cohort 26 were log2 transformed to minimize outliers. Next, Wnt target genes were selected based on their names and alias terms as described in paragraph 2.4. In case of multiple probes representing a single gene, the probe with the highest variability, based on the maximum IQR, using the entire dataset, was selected for further analyses. An unsupervised hierarchical cluster analysis was performed using median centred expression data with Cluster 3.0 for Windows (Human Genome Center, University of Tokyo, 2002) and Java Treeview, version 1.1.6rv, for visualization. The clustering was based on the centred correlation as a distance metric using average linkage. Ethical approval This study was part of a protocol entitled “Translational research on soft tissue sarcomas” which was assessed by the Medical Ethics Committee of the Erasmus MC, Rotterdam, the Netherlands (MEC-2016-213). 3