TY - JOUR
T1 - A pair of new statistical parameters for quality control in RNA interference high-throughput screening assays
AU - Zhang, Xiaohua Douglas
PY - 2007/4
Y1 - 2007/4
N2 - RNA interference (RNAi) high-throughput screening (HTS) enables massive parallel gene silencing and is increasingly being used to reveal novel connections between genes and disease-relevant phenotypes. The application of genome-scale RNAi relies on the development of high-quality RNAi HTS assays. To obtain high-quality HTS assays, there is a strong need for an easily interpretable and theoretically based quality control (QC) metric. Signal-to-noise ratio (S/N), signal-to-background ratio (S/B), and Z-factor have been adopted as QC metrics in HTS assays. In this paper, I proposed a pair of new parameters, strictly standardized mean difference (SSMD) and coefficient of variability in difference (CVD), as QC metrics in RNAi HTS assays. Compared to S/B and S/N, SSMD and CVD capture the variabilities in both compared populations. Compared to Z-factor, SSMD and CVD have a clear probability interpretation and a solid statistical basis. Accordingly, the cutoff criteria of using SSMD or CVD as a QC metric in HTS assays are fully theoretically based. In addition, I discuss the relationship between the SSMD-based criterion and the popular Z-factor-based criterion and elucidate why p-value from t-test of testing mean difference fails to serve as a QC metric.
AB - RNA interference (RNAi) high-throughput screening (HTS) enables massive parallel gene silencing and is increasingly being used to reveal novel connections between genes and disease-relevant phenotypes. The application of genome-scale RNAi relies on the development of high-quality RNAi HTS assays. To obtain high-quality HTS assays, there is a strong need for an easily interpretable and theoretically based quality control (QC) metric. Signal-to-noise ratio (S/N), signal-to-background ratio (S/B), and Z-factor have been adopted as QC metrics in HTS assays. In this paper, I proposed a pair of new parameters, strictly standardized mean difference (SSMD) and coefficient of variability in difference (CVD), as QC metrics in RNAi HTS assays. Compared to S/B and S/N, SSMD and CVD capture the variabilities in both compared populations. Compared to Z-factor, SSMD and CVD have a clear probability interpretation and a solid statistical basis. Accordingly, the cutoff criteria of using SSMD or CVD as a QC metric in HTS assays are fully theoretically based. In addition, I discuss the relationship between the SSMD-based criterion and the popular Z-factor-based criterion and elucidate why p-value from t-test of testing mean difference fails to serve as a QC metric.
KW - Coefficient of variability in difference
KW - High-throughput screening
KW - RNA interference
KW - Strictly standardized mean difference
KW - Z-factor
KW - d-probability
UR - http://www.scopus.com/inward/record.url?scp=33947167826&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=33947167826&partnerID=8YFLogxK
U2 - 10.1016/j.ygeno.2006.12.014
DO - 10.1016/j.ygeno.2006.12.014
M3 - Article
C2 - 17276655
AN - SCOPUS:33947167826
SN - 0888-7543
VL - 89
SP - 552
EP - 561
JO - Genomics
JF - Genomics
IS - 4
ER -