“…Although many imputation methods exist, we considered the most conservative approach was to use the available data without imputation as it is often more robust and is less likely to suffer from any other biases [17]. Many of these 16 proteins have previously been shown to influence metastasis in other cancers, for instance, ACTG1 in liver cancer [18], Immunoglobulin heavy constant gamma 3 (IGHG3) in breast cancer [19], Proteasome subunit alpha type-1 (PSMA1) in colon cancer [20], Ras-related protein Rab-11A (RAB11a) in pancreatic cancer [21], PEBP1 and Eukaryotic translation elongation factor 1 alpha 1 (EEF1A1) in hepatocellular carcinoma [22,23], MYH9 in colorectal [24] and ovarian cancer [25], POSTN in melanoma [26], hepatocellular carcinoma [27] and breast cancer [28] and SND1 in breast cancer [29] and cervical cancer [30]. Furthermore, five of these proteins were identified as significantly differentially expressed between PM and PNMs in two NCBI GEO datasets and the gene expression of 6 of these 16 significantly differentially expressed proteins (i.e., EVPL, POSTN, KRT9, MYH9, MYH16 and PEBP1) each had a significant effect on 3-year survival rates in cutaneous melanoma as identified from the TCGA database.…”