Figure_3.tif (1.12 MB)

Circuit diagram representing the probability of various outcomes in two-stage RNAi screening.

Download (0 kB)
posted on 2013-09-19, 02:40 authored by Linhui Hao, Qiuling He, Zhishi Wang, Mark Craven, Michael A. Newton, Paul Ahlquist

A) Detection Screen: The diagram expresses ways in which treated cells (left) can progress through the experiment to a scored detection Dg,s, conditional upon the state of latent variables (accessibility Ag,s, involvement Ig, and off target count Tg,s). In the case shown, cells are treated with a pool of 4 siRNAs against a single gene g. The top branch involves a phenotypically effective knockdown event that blocks influenza virus infection, due to an on-target effect (upper sub-branch) or off-target event (lower sub-branch) or both. The alternate bottom branch involves the absence of siRNA-mediated interference with influenza virus replication. In either case, as shown, negative measurement error (type II) or positive measurement error (type I) could affect the final scored phenotype. Depending on these combined effects, the gene g in influenza virus replication is scored as detected (1) or not detected (0). Shaded boxes record the probability of the indicated event, where variables a and i are both either 0 or 1, and count t is a natural number. Accordingly, phenotype probabilities are computed by adding probabilities over all paths from the left to the specific outcome, where a path probability is computed by multiplying over the traversed events. The open box at the bottom provides a concise summary form for the conditional detection probability. B) Schematic similar to (A) regarding the possible outcomes and associated probabilities for secondary confirmation testing of a given gene g implicated by detection screening in study s. Cells are treated in four separate assays with each of the four distinct siRNAs targeting gene g, and can then traverse one of the two main indicated outcome branches, similar to panel (A). As noted, confirmation requires a positive result with at least two of the individual siRNAs against gene g.


Usage metrics

    PLOS Computational Biology



    Ref. manager