MATERIAL AND METHODS

HBV gene signature

An HBV cccDNA destabilizing gene signature was extracted from an external micro array data set (GEO:GSE46667) \cite{Lucifora_2013}. In this experiment, HBV infected, differentiated HepaRG cells were either treated or not with a dHepaRG antibody. This antibody is know to interact with the lymphotoxin beta receptor which results in the induction of the cytidine deaminase APOBEC3B. Once induced, APOBEC3B will destabilize the cccDNA of HBV \cite{Lucifora_2013} and as such eliminate the viral genome from the host. Transcriptional differences between four biological replicates of the antibody treated and four biological replicates of the non treated HBV infected differentiated HepaRG cells were assessed via a  Limma differential gene expression analysis \cite{Ritchie_2015}. The p-values were moderated across all tested genes. The eBayes function of the R limma package was used for this purpose at a 5% significance level. The Benjamini-Hochberg algorithm was applied to correct for multiple testing. 

L1000 assay

A total of 257402 samples from 237390 unique compounds were transcriptionally profiled in the MCF7 human breast epithelial adenocarcinoma cell line (ATCC HTB-22). After the start of each new cell culture, one flask of MCF7 cells was tested for possible mycoplasma contamination. 5000 MCF7 cells were plated per well in a 384-well cell culture plate (i.e. 40µL/well). The cells were incubated for 24 hours at 37°C. When the monolayer was approximately 75% confluent, 1µM of compound, positive controls (i.e.LY-294002 n=710, Estradiol n=730, Monorden n=617, Sirolimus n=107, Trichostatin n=729) or vehicle samples (i.e. DMSO) were added to a final volume of 50µL/well. The 384 well plates were stored in an incubator for seven hours at 37°C, after which 45µL of a lysis buffer (i.e. Qiagen, 1031576) was added to each well. The plates were subsequently kept at room temperature for 30’ and sealed after which they were stored at -80°C. The lysates were shipped on dry ice to Genometry, Inc (Cambridge, MA, US), where the L1000 assay is ran. The data were preprocessed by Genometry, Inc (Cambridge, MA, US), following their standard procedure of gene assignment, invariant set normalisation within each sample, quantile normalisation across samples within each 384-well plate and a log2 transformation of the signal. 

L1000 based transcriptional similarity 

The transcriptional effects for each of the 257402 samples, were assessed via a Limma differential gene expression analysis \cite{Ritchie_2015}. The t-and p-values from this test were computed on the log2 transformed signal for each of the 978 measured genes versus the vehicle sample distribution within the same plate. The p-values were moderated across all tested genes, with the variances of the vehicle samples pooled across all genes. The eBayes function of the R limma package was used for this purpose at a 5% significance level. The Benjamini-Hochberg algorithm was applied to correct for multiple testing. As such, each compound could be re-written as a vector of t- and p-statistics. The magnitude of the compound effect is given by the p-value while the direction of the effect is denoted by the sign of the t-value. Based on these two parameters, all 978 genes could be ranked and signed for each of the 257402 samples.  Transcriptional similarity between the HBV gene signature and a transcriptional profile of a compound was computed using the standardized ranked Zhang connection score c that ranges between +1 and -1 \cite{Zhang_2008}. Confidence intervals were computed for the standardized Zhang connection scores, setting the false positive rate at 1, using the methodology described by De Wolf et al (2018, submitted).

Dose response analysis of L1000 selected compounds

Albumine
t-HBV DNA (extracellular)
HBV-RNA (intracellular)
HBeAg (Elisa)
HBsAg (Alphalisa)
ANDERE TESTEN IN DEZE CONTEXT, CONFIRMATIE EXPERIMENTEN DIE AANLEIDING GEVEN TOT HET FEIT DAT DE GEKOZEN CPD(S) IETS DOEN MET HBV, MAAR OOK DAT HET ENTRY PROCESS KAN WORDEN AANGEDUID ALS MOA

Time to addition of HIT compound