Generation of an shp53 epi-allelic series. (a) HL60 p53-RNAi reporter cell line. A HL60 cell line was transduced with p53-RNAi reporter construct and a single cell-derived clone was isolated by FACS, expanded and reanalysed for optimal (GFP) reporter expression by flow cytometry. (b) HL60 p53-RNAi reporter cells transduced with shp53 library show a population with reduced RNAi reporter expression. HL60 p53-RNAi reporter cells were transduced with a library of candidate p53 shRNA or control shRNA (shLuc). All cells were monitored by FACS and cells with reporter levels within the GFPlow/RFP + gate (box) were sorted. (c) Functionally selected shRNA sequence frequency profile. Sequences retrieved from the 454 deep sequencing were tested to determine whether they encoded a hairpin structure and clustered into groups of near-identical sequences, and plotted by frequency. (d) Generalized linear model (GLM) fit of shRNA sequence frequency versus RNAi reporter level reveals a statistical relationship. Representative shp53-sequences were recloned, transduced into HL60 p53-RNAi reporter cells and analyzed by flow cytometry. A generalized linear model (GLM) with a log link function (‘log(frequency) ~ 1 + knockdown’) and Poisson distribution was applied to model the shRNA frequency versus RNAi reporter fluorescence intensity, generating estimated coefficients listed in Table 2, Additional file 4: Figure S4. The GLM model shows an exponential relationship between the observed shRNA frequency and measured RNAi reporter at a significance level of α = 0.05 (p ≈ 0). The reporter intensity data represents the mean of three independent experiments. (e) HUVE cells transduced with different shp53 clones generates an epi-allelic series. Transduced HUVE cells expressing different shp53 shRNA sequences show graded p53 expression levels as measured by relative quantification of Western blot versus mRNA levels.