Limits...
Exploring the molecular causes of hepatitis B virus vaccination response: an approach with epigenomic and transcriptomic data.

Lu Y, Cheng Y, Yan W, Nardini C - BMC Med Genomics (2014)

Bottom Line: Twenty-five infants were recruited and classified as good and non-/low- responders according to serological test results.Results were finally associated to already published transcriptomics and post-transcriptomics to gain further insight.Future research in this direction is warranted.

View Article: PubMed Central - HTML - PubMed

Affiliation: Group of Clinical Genomic Networks, Key Laboratory of Computational Biology, CAS-MPG, Partner Institute for Computational Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, PR China. yanwl@fudan.edu.cn.

ABSTRACT

Background: Variable responses to the Hepatitis B Virus (HBV) vaccine have recently been reported as strongly dependent on genetic causes. Yet, the details on such mechanisms of action are still unknown. In parallel, altered DNA methylation states have been uncovered as important contributors to a variety of health conditions. However, methodologies for the analysis of such high-throughput data (epigenomic), especially from the computational point of view, still lack of a gold standard, mostly due to the intrinsic statistical distribution of methylomic data i.e. binomial rather than (pseudo-) normal, which characterizes the better known transcriptomic data.We present in this article our contribution to the challenge of epigenomic data analysis with application to the variable response to the Hepatitis B virus (HBV) vaccine and its most lethal degeneration: hepatocellular carcinoma (HCC).

Methods: Twenty-five infants were recruited and classified as good and non-/low- responders according to serological test results. Whole genome DNA methylation states were profiled by Illumina HumanMethylation 450 K beadchips. Data were processed through quality and dispersion filtering and with differential methylation analysis based on a combination of average methylation differences and non-parametric statistical tests. Results were finally associated to already published transcriptomics and post-transcriptomics to gain further insight.

Results: We highlight 2 relevant variations in poor-responders to HBV vaccination: the hypomethylation of RNF39 (Ring Finger Protein 39) and the complex biochemical alteration on SULF2 via hypermethylation, down-regulation and post-transcriptional control.

Conclusions: Our approach appears to cope with the new challenges implied by methylomic data distribution to warrant a robust ranking of candidates. In particular, being RNF39 within the Major Histocompatibility Complex (MHC) class I region, its altered methylation state fits with an altered immune reaction compatible with poor responsiveness to vaccination. Additionally, despite SULF2 having been indicated as a potential target for HCC therapy, we can recommend that non-responders to HBV vaccine who develop HCC are quickly directed to other therapies, as SULF2 appears to be already under multiple molecular controls in such patients. Future research in this direction is warranted.

Show MeSH

Related in: MedlinePlus

MDS plot. Multiple dimensional scaling (MDS) plot of the 1000 most variable loci showing that no significant batch effect can be detected, while samples genders are easily discriminated.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4008305&req=5

Figure 1: MDS plot. Multiple dimensional scaling (MDS) plot of the 1000 most variable loci showing that no significant batch effect can be detected, while samples genders are easily discriminated.

Mentions: Quality Control (QC) assessment was performed with the open-source R package minfi[13]. Data distribution and intensity of internal control probes including bisulfite I, II, hybridization, extension, specificity I, II, target removal were checked and no major defects were spotted for the QC. To evaluate the presence of any batch effect, we performed multiple dimensional scaling (MDS), a dimensional reduction approach to visualize the distances (similarities) of individual cases in a dataset, using the function mdsPlot in the package “minfi”, on the 1000 most variable positions of the merged raw data. No significant batch effect was detected while the samples’ genders were well discriminated (see Figure 1). Basic quality filtering was then performed to the control-normalized and background-subtracted data exported from Illumina software GenomeStudio. Stringent data filtering was done according to recent recommendations [14] to control statistical power and reduce false discoveries. In particular, loci with detection p-value > 0.01 were removed along with loci having more than 20% NAs (number of detecting beads < 3) in control or case. Potential confounding factors were additionally controlled by removal of the X/Y chromosomes. Dispersion filtering measured by standard deviation (SD) and interquantile range (IQR) with cutoffs set to remove 80% of the least varying loci [14] was also performed. To achieve stringent filtering, at this stage, the intersection of the results of the 2 metrics (SD and IQR) was preserved, overall reducing the candidate loci list from ~450,000 (485,577) to 76,074.


Exploring the molecular causes of hepatitis B virus vaccination response: an approach with epigenomic and transcriptomic data.

Lu Y, Cheng Y, Yan W, Nardini C - BMC Med Genomics (2014)

MDS plot. Multiple dimensional scaling (MDS) plot of the 1000 most variable loci showing that no significant batch effect can be detected, while samples genders are easily discriminated.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4008305&req=5

Figure 1: MDS plot. Multiple dimensional scaling (MDS) plot of the 1000 most variable loci showing that no significant batch effect can be detected, while samples genders are easily discriminated.
Mentions: Quality Control (QC) assessment was performed with the open-source R package minfi[13]. Data distribution and intensity of internal control probes including bisulfite I, II, hybridization, extension, specificity I, II, target removal were checked and no major defects were spotted for the QC. To evaluate the presence of any batch effect, we performed multiple dimensional scaling (MDS), a dimensional reduction approach to visualize the distances (similarities) of individual cases in a dataset, using the function mdsPlot in the package “minfi”, on the 1000 most variable positions of the merged raw data. No significant batch effect was detected while the samples’ genders were well discriminated (see Figure 1). Basic quality filtering was then performed to the control-normalized and background-subtracted data exported from Illumina software GenomeStudio. Stringent data filtering was done according to recent recommendations [14] to control statistical power and reduce false discoveries. In particular, loci with detection p-value > 0.01 were removed along with loci having more than 20% NAs (number of detecting beads < 3) in control or case. Potential confounding factors were additionally controlled by removal of the X/Y chromosomes. Dispersion filtering measured by standard deviation (SD) and interquantile range (IQR) with cutoffs set to remove 80% of the least varying loci [14] was also performed. To achieve stringent filtering, at this stage, the intersection of the results of the 2 metrics (SD and IQR) was preserved, overall reducing the candidate loci list from ~450,000 (485,577) to 76,074.

Bottom Line: Twenty-five infants were recruited and classified as good and non-/low- responders according to serological test results.Results were finally associated to already published transcriptomics and post-transcriptomics to gain further insight.Future research in this direction is warranted.

View Article: PubMed Central - HTML - PubMed

Affiliation: Group of Clinical Genomic Networks, Key Laboratory of Computational Biology, CAS-MPG, Partner Institute for Computational Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, PR China. yanwl@fudan.edu.cn.

ABSTRACT

Background: Variable responses to the Hepatitis B Virus (HBV) vaccine have recently been reported as strongly dependent on genetic causes. Yet, the details on such mechanisms of action are still unknown. In parallel, altered DNA methylation states have been uncovered as important contributors to a variety of health conditions. However, methodologies for the analysis of such high-throughput data (epigenomic), especially from the computational point of view, still lack of a gold standard, mostly due to the intrinsic statistical distribution of methylomic data i.e. binomial rather than (pseudo-) normal, which characterizes the better known transcriptomic data.We present in this article our contribution to the challenge of epigenomic data analysis with application to the variable response to the Hepatitis B virus (HBV) vaccine and its most lethal degeneration: hepatocellular carcinoma (HCC).

Methods: Twenty-five infants were recruited and classified as good and non-/low- responders according to serological test results. Whole genome DNA methylation states were profiled by Illumina HumanMethylation 450 K beadchips. Data were processed through quality and dispersion filtering and with differential methylation analysis based on a combination of average methylation differences and non-parametric statistical tests. Results were finally associated to already published transcriptomics and post-transcriptomics to gain further insight.

Results: We highlight 2 relevant variations in poor-responders to HBV vaccination: the hypomethylation of RNF39 (Ring Finger Protein 39) and the complex biochemical alteration on SULF2 via hypermethylation, down-regulation and post-transcriptional control.

Conclusions: Our approach appears to cope with the new challenges implied by methylomic data distribution to warrant a robust ranking of candidates. In particular, being RNF39 within the Major Histocompatibility Complex (MHC) class I region, its altered methylation state fits with an altered immune reaction compatible with poor responsiveness to vaccination. Additionally, despite SULF2 having been indicated as a potential target for HCC therapy, we can recommend that non-responders to HBV vaccine who develop HCC are quickly directed to other therapies, as SULF2 appears to be already under multiple molecular controls in such patients. Future research in this direction is warranted.

Show MeSH
Related in: MedlinePlus