28.12.2013 Views

genomewide characterization of host-pathogen interactions by ...

genomewide characterization of host-pathogen interactions by ...

genomewide characterization of host-pathogen interactions by ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Maren Depke<br />

Results<br />

Kidney Gene Expression Pattern in an in vivo Infection Model<br />

infection with<br />

RN1HG<br />

BR1<br />

(4 arrays)<br />

3<br />

infection with<br />

RN1HG<br />

BR2<br />

(5 arrays)<br />

4<br />

2<br />

1<br />

2<br />

sham infection<br />

(NaCl)<br />

BR2<br />

(5 arrays)<br />

infection with<br />

RN1HG ΔsigB<br />

BR1<br />

(5 arrays)<br />

3<br />

infection with<br />

RN1HG ΔsigB<br />

BR2<br />

(5 arrays)<br />

4<br />

Fig. R.2.3: Overview on the comparisons between groups in this study that were addressed with statistical testing and visualized with<br />

scatter plots.<br />

The two groups with different infecting strains (infection with RN1HG ΔsigB vs. infection with RN1HG) were compared after<br />

combining both biological replicates () and for each biological replicate separately (). Same experimental groups <strong>of</strong> different<br />

biological replicates (infection with RN1HG: BR1 vs. BR2; infection with RN1HG ΔsigB: BR1 vs. BR2) were also checked against each<br />

other (). The last comparison focused on infection and sham infection in the second biological replicate (infection with RN1HG vs.<br />

sham infection; infection with RN1HG ΔsigB vs. sham infection; ).<br />

BR – biological replicate.<br />

The comparisons were visualized using scatter plots (Fig. R.2.4). Comparable to the PCA<br />

results, a striking concordance between expression values <strong>of</strong> kidney after infection with RN1HG<br />

and infection with RN1HG ΔsigB was observed, especially when both biological replicates were<br />

combined (Fig. R.2.4, panel ), but also when the biological replicates were considered<br />

separately (Fig. R.2.4, panel ). In the comparison <strong>of</strong> the same experimental groups in both<br />

replicates high similarity was observed (Fig. R.2.4, panel ), although the inter-replicate variation<br />

<strong>of</strong> the same treatment was higher than the intra-replicate variation <strong>of</strong> the different treatments<br />

for S. aureus infected samples. The inter-replicate variation might be due to the difference <strong>of</strong> one<br />

day in the sampling time point (d 4 or d 5). Strong effects <strong>of</strong> S. aureus infection independent <strong>of</strong><br />

the infecting strain emerged in the comparison to the sham infected / NaCl control sample group<br />

(Fig. R.2.4, panel ).<br />

The different experimental groups were compared with statistical testing to obtain lists <strong>of</strong><br />

differentially expressed genes. Sequences not expressed and control sequences were not<br />

included in statistical testing.<br />

The first statistical test compared kidney samples <strong>of</strong> mice infected with S. aureus RN1HG ΔsigB<br />

vs. infection with RN1HG in an approach that combined both biological replicates (for<br />

comparison see Fig. R.2.3; and Fig. R.2.4, panel ). Only one sequence corresponding to one<br />

gene was significantly different in intensity between both groups (Table R.2.1). For all arrays<br />

except one array from a specific animal the signal intensities <strong>of</strong> this gene were low and their p-<br />

value for expression was high. i. e. the expression was absent (Fig. R.2.5 A). The array data from<br />

the single outlying animal caused statistical significance, but the result is not biologically relevant<br />

because it is obviously due to an unknown, animal-specific factor and not to treatment.<br />

77

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!