S4 object storing the results of SAINTq: the SAINTq
interaction-scoring tables, the corresponding volcano plots, and the
parameters used to compute the scores.
scoresA named list (one element per bait) of data.frames
containing the SAINTq scoring tables. Each table reports one row per
bait-prey pair with the following columns:
Bait: name of the bait group (a value of the metadata.column) scored in this table; constant within each per-bait element of the list.
Control: name of the control group (a value of the metadata.column).
Prey: identifier of the prey protein (the row name of the LFQ count matrix).
n.rep: number of bait replicates in which the prey was quantified (i.e. non-missing values).
AvgP: average probability, the main SAINTq score.
It is the posterior probability of a true (bait-specific) interaction
averaged across the bait replicates. Ranges in [0, 1]; higher
values indicate higher confidence.
MaxP: the maximum per-replicate posterior probability across the bait replicates (the score of the single best replicate).
log2.FoldChange_bait.vs.control: enrichment of the prey in the bait over the control, on the linear scale, computed as log2(avg.bait) - log2(avg.ctrl).
avg.bait: mean log2 intensity of the prey across the bait replicates.
avg.ctrl: mean log2 intensity of the prey across the control run (the background mean \(\mu_F\) of the model;
replaced by the global background value for preys never detected in the control).
bFDR: Bayesian false discovery rate of the interaction,
derived from the AvgP values as the cumulative mean of (1 - AvgP)
down the probability-ranked list. Lower values indicate higher confidence.
Class: "ANY".
volcanoesA named list (one element per bait) of ggplot
objects. Each is a volcano-style plot of the interactions for that bait,
showing the log2(FoldChange) on the x-axis against -log10(bFDR)
on the y-axis, with the size and color of the points encoding the
AvgP score. Class: "ANY".
parametersA list recording the settings and fitted quantities used for the scoring:
prior.pi: prior probability of a true interaction (\(\pi_T\));
either estimated by expectation-maximization or fixed by the user.
fold: fold change separating the true component from the
background, such that mu_T = mu_F + log2(fold).
delta.log2: log2(fold), i.e. the additive shift (in log2 units)
applied to the background mean to define the mean of the true component.
sd.scale: scaling factor of the true-component standard
deviation relative to the background one (sigma_T = sd.scale * sigma_F).
min.sd: lower bound applied to the per-prey background standard deviation,
preventing over-confident scores from preys with near-constant control intensities.
sigma.global: global background standard deviation (median of the per-prey control standard deviations),
used as a fallback for preys with too few control measurements.
background: log2 background intensity assigned to preys never detected in any control run.
which.data: the DEprot count matrix that was scored, one of "imputed", "randomized", "normalized" or "raw".
control: name of the control group used as the background.
baits: name(s) of the bait group(s) that were scored.
best.n.rep: number of top-scoring bait replicates used when averaging the per-replicate posterior probabilities
into AvgP (the "best R replicates" option of SAINTexpress). NULL means that all replicates of each bait were used.
Class: "ANY".