Comparison of our study with datasets on nuclear subcompartments. Datasets were created for each nuclear subcompartment based on online databases or recent review articles. NOPdb (Ahmad et al., 2009) was used to represent nucleolar proteins. A proteomic analysis of interchromatin granule clusters (Saitoh et al., 2004) was used to represent the nuclear speckles dataset. A PML body interactome analysis (Van Damme et al., 2010) was used to represent proteins in PML bodies. A list of Cajal body proteins from a recent review paper (Machyna et al., 2013) was used to represent proteins in Cajal bodies. A list of paraspeckle proteins from a review (Bond and Fox, 2009) was used to represent proteins in paraspeckles. Venn graphs were used to show the extent of overlapping between an available dataset (green) and our study (blue). A group of proteins that are uniquely identified in our study and have been reported by others in the literature were presented in dark blue. The other groups of proteins that are confirmed by our shRNA screen to be involved in the assembly of Cajal bodies or paraspeckles were presented in yellow. Please also see Table S2 and Table S3.