Skip to main content

Table 1.

Likelihood values for Bayes' analysis

Dataset Criterion P
1 Transcriptome, mouse mpkCCD median normalized abundance >1 0.95
1 > median normalized abundance > 0.6 0.70
0.6 > median normalized abundance 0.10
2 Proteome, mouse mpkCCD detected in proteome 0.95
not detected in proteome 0.50
3 Dot product with AQP2, mouse mpkCCD* MAD normalized value >50 0.90
5 > MAD normalized value > 50 0.70
1 > MAD normalized value > 5 0.60
0 > MAD normalized value > 1 0.40
MAD normalized value >1 when deviation is negative 0.20
4 RNA-sequence, rat** RPKM values >1.0 in CNT, CCD, OMCD, and IMCD 0.80
RPKM values >1.0 in three of the four segments 0.60
RPKM values >1.0 in two of the four segments 0.40
RPKM values >1.0 in one or none of the four segments 0.20
5 UniProt subcellular location nucleus only 0.20
cytoplasm/cytosol/membrane/endosome 0.80
6 UniProt domain/IMP IMP 0.80
in a transmembrane protein-interacting domain 0.80
all others 0.50

Criterion and probabilities used in Bayes' analysis. If there is no information on an E3 ubiquitin ligase, it received a probability P = 0.5 (coin toss). To see the rationale for these values, see Supplemental Methods.

*

A dot product takes vectors/arrays of numbers and returns a single value. To do this, we multiply the protein abundance levels for aquaporin-2 and the E3 ligase in each fraction and then sum the products over all of the fractions to end with a single value. We created dot products between aquaporin-2 and each of the E3 ubiquitin ligases found in the differential centrifugation fractions of cultured mouse mpkCCD cells. These dot products were used to assign probabilities to each E3 ligase based on their colocalization with aquaporin-2. To create these probabilities, we used median absolute deviation (MAD), which is defined as the median of the absolute deviations from the median (11). We found the median absolute deviation, and then divided each ligase's absolute deviation from the median by it.

**

Aquaporin-2 is found primarily in the renal rat connecting tubule (CNT), cortical collecting duct (CCD), outer medullary collecting duct (OMCD), and inner medullary collecting duct (IMCD). E3 ligases were given probabilities dependent on how many of these tubule segments they were found in.