MultiplexMixer sorts samples into multiplex “channel” assignments to mitigate potential sample bias and batch effect for cohort scale quantitative mass spectrometry (MS)-based proteomic workflows. MultiplexMixer intakes a sample dataset file and attempts to homogenize sample attributes across multiplexes according to user-assigned variable weighting and isobaric channel availability. Following sample sorting, MultiplexMixer exports metrics and visualizations to illustrate the degree of homogeneity for sample attribute distribution across multiplex assignments.
NOTE: There is a minor aspect of randomness to MultiplexMixer in respect to sample sorting. As such, the same input could yield minor differences in results.
For any questions concerning MultiplexMixer, please contact either:
Dr. Nicholas W. Bateman, Women’s Health Integrated Research Center at Inova Health System, 3289 Woodburn Road, Suite 370, Annandale, VA 22003. (
batemann@whirc.org
)
Dr. Thomas P. Conrads, 3289 Woodburn Rd., Suite 375, Annandale, VA 22003. (
conrads@whirc.org
)
Chi-sq test of homogeneity, measures the degree to which species for each variables of interest were able to be evenly distributed amongst the available plex channels. Since homogenous distribution is the goal, an optimal p-value is defined as: ‘p-val >= 0.95’. Standard p-value for this test of homogeneity (p <= 0.05) indicates that the group of species for the variable under consideration for at least one of the plexes differs significantly from the grouping of species in other plexes.
JSI Heatmap for variables:
MultiplexMixer calculates Jaccard Similarity Index (JSI)
(Jaccard 1901)(Jaccard 1912)
for pairwise plex comparisons to measure the similarity of variable species distribution across plexes. Comparisons of sample attribute likeness is determined on scale between 0 (dissimilar) and 1 (similar).
JSI Comparison p-values for variables:
JSI are plotted against p-value curves
(Real 1999)(Béguinot 2019)
which are determine by the sum of the species in two occupational units.
Effective Number Species for variables:
The below boxplots illustrate distributions of Species Richness and Effective Number Species
(MACARTHUR 1965)
calculations from Shannon Entropy
(Shannon 1948)(Shannon 1948)
and Simpson's Diversity Index
(Simpson 1949)
for variable attributes across plexes. Species Richness defines the number of unique species in a plex. Effective Number Species calculations from diversity indices determine the true number of species in a plex.
Table with the top ten most frequent co-occuring species per plex
The table column 'n' displays the frequency for the listed species for the listed plex--Meaning that there are 'n' samples that share the listed species for the variables of interest in the plex.
Barplot of plex assignments for variables:
Barplots illustrate the species' assignments for each variable across plexes