Limits...
Impact of RNA degradation on gene expression profiling.

Opitz L, Salinas-Riester G, Grade M, Jung K, Jo P, Emons G, Ghadimi BM, Beissbarth T, Gaedcke J - BMC Med Genomics (2010)

Bottom Line: Only a relatively small number of probes (275 out of 41,000) show a significant effect due to degradation.A much higher biological variance between patients is observed compared to the effect that is imposed by degradation of RNA.These results are limited to the Agilent 44 k microarray platform and should be carefully interpreted when transferring to other settings.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department Medical Statistics, University Medicine Göttingen, Humboldtallee 32, 37073 Göttingen, Germany.

ABSTRACT

Background: Gene expression profiling is a highly sensitive technique which is used for profiling tumor samples for medical prognosis. RNA quality and degradation influence the analysis results of gene expression profiles. The impact of this influence on the profiles and its medical impact is not fully understood. As patient samples are very valuable for clinical studies, it is necessary to establish criteria for the RNA quality to be able to use these samples in later analysis.

Methods: To investigate the effects of RNA integrity on gene expression profiling, whole genome expression arrays were used. We used tumor biopsies from patients diagnosed with locally advanced rectal cancer. To simulate degradation, the isolated total RNA of all patients was subjected to heat-induced degradation in a time-dependent manner. Expression profiling was then performed and data were analyzed bioinformatically to assess the differences.

Results: The differences introduced by RNA degradation were largely outweighed by the biological differences between the patients. Only a relatively small number of probes (275 out of 41,000) show a significant effect due to degradation. The genes that show the strongest effect due to RNA degradation were, especially, those with short mRNAs and probe positions near the 5' end.

Conclusions: Degraded RNA from tumor samples (RIN > 5) can still be used to perform gene expression analysis. A much higher biological variance between patients is observed compared to the effect that is imposed by degradation of RNA. Nevertheless there are genes, very short ones and those with the probe binding side close to the 5' end that should be excluded from gene expression analysis when working with degraded RNA. These results are limited to the Agilent 44 k microarray platform and should be carefully interpreted when transferring to other settings.

Show MeSH

Related in: MedlinePlus

Effect of RNA degradation on comparability between patients. (A) Two-dimensional PCA plot of genome-wide expression profiles showing principal components 1 and 2. The first axis (PC1) accounts for 33% of the overall variance of the data, the second axis accounts for 29% (PC2). The colors blue (P159), red (P160) and green (P162) refer to the different patients. The degradation levels are represented by the following symbols: C (0:00 h), 1 (1:45 h), 2 (2:30 h) and 3 (3:15 h). (B) shows pairwise correlations between all samples of patients. The cells in the visualization are colored by Pearson's correlation coefficient values with deeper colors indicating higher positive (blue) correlations. The heatmap is flanked by clustering dendrograms showing the similarity between samples in a hierarchical approach.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2927474&req=5

Figure 2: Effect of RNA degradation on comparability between patients. (A) Two-dimensional PCA plot of genome-wide expression profiles showing principal components 1 and 2. The first axis (PC1) accounts for 33% of the overall variance of the data, the second axis accounts for 29% (PC2). The colors blue (P159), red (P160) and green (P162) refer to the different patients. The degradation levels are represented by the following symbols: C (0:00 h), 1 (1:45 h), 2 (2:30 h) and 3 (3:15 h). (B) shows pairwise correlations between all samples of patients. The cells in the visualization are colored by Pearson's correlation coefficient values with deeper colors indicating higher positive (blue) correlations. The heatmap is flanked by clustering dendrograms showing the similarity between samples in a hierarchical approach.

Mentions: Next, we analyzed the samples of the different patients at different RNA degradation levels using gene expression profiling. With this we address the question of how significant the influence of RNA degradation is on the gene expression profiles in comparison to the overall influence of different patients. To gain insights into the data obtained by the microarrays and, especially, to analyze how patient samples correlate with degradation stages, different bioinformatic methods were applied. The multivariate data analysis method PCA (principal components analysis) was used to visualize similarities or dissimilarities between genome-wide expression profiles of patient samples and degradation stages in a two-dimensional plot. PCA is a linear projection method that allows visualization of high-dimensional data in a lower dimensional space. The results of the PCA analysis for the normalized microarray dataset are shown in Figure 2A. The first axis of the plot (first principal component, PC1) accounts for 33% of the overall variance of the data, while the second axis accounts for 29% (PC2). It can clearly be seen that samples from the same patient are very similar to each other regardless of the degradation level. The differences between the different patients, which are to be analyzed, contribute the major part of the overall differences.


Impact of RNA degradation on gene expression profiling.

Opitz L, Salinas-Riester G, Grade M, Jung K, Jo P, Emons G, Ghadimi BM, Beissbarth T, Gaedcke J - BMC Med Genomics (2010)

Effect of RNA degradation on comparability between patients. (A) Two-dimensional PCA plot of genome-wide expression profiles showing principal components 1 and 2. The first axis (PC1) accounts for 33% of the overall variance of the data, the second axis accounts for 29% (PC2). The colors blue (P159), red (P160) and green (P162) refer to the different patients. The degradation levels are represented by the following symbols: C (0:00 h), 1 (1:45 h), 2 (2:30 h) and 3 (3:15 h). (B) shows pairwise correlations between all samples of patients. The cells in the visualization are colored by Pearson's correlation coefficient values with deeper colors indicating higher positive (blue) correlations. The heatmap is flanked by clustering dendrograms showing the similarity between samples in a hierarchical approach.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2927474&req=5

Figure 2: Effect of RNA degradation on comparability between patients. (A) Two-dimensional PCA plot of genome-wide expression profiles showing principal components 1 and 2. The first axis (PC1) accounts for 33% of the overall variance of the data, the second axis accounts for 29% (PC2). The colors blue (P159), red (P160) and green (P162) refer to the different patients. The degradation levels are represented by the following symbols: C (0:00 h), 1 (1:45 h), 2 (2:30 h) and 3 (3:15 h). (B) shows pairwise correlations between all samples of patients. The cells in the visualization are colored by Pearson's correlation coefficient values with deeper colors indicating higher positive (blue) correlations. The heatmap is flanked by clustering dendrograms showing the similarity between samples in a hierarchical approach.
Mentions: Next, we analyzed the samples of the different patients at different RNA degradation levels using gene expression profiling. With this we address the question of how significant the influence of RNA degradation is on the gene expression profiles in comparison to the overall influence of different patients. To gain insights into the data obtained by the microarrays and, especially, to analyze how patient samples correlate with degradation stages, different bioinformatic methods were applied. The multivariate data analysis method PCA (principal components analysis) was used to visualize similarities or dissimilarities between genome-wide expression profiles of patient samples and degradation stages in a two-dimensional plot. PCA is a linear projection method that allows visualization of high-dimensional data in a lower dimensional space. The results of the PCA analysis for the normalized microarray dataset are shown in Figure 2A. The first axis of the plot (first principal component, PC1) accounts for 33% of the overall variance of the data, while the second axis accounts for 29% (PC2). It can clearly be seen that samples from the same patient are very similar to each other regardless of the degradation level. The differences between the different patients, which are to be analyzed, contribute the major part of the overall differences.

Bottom Line: Only a relatively small number of probes (275 out of 41,000) show a significant effect due to degradation.A much higher biological variance between patients is observed compared to the effect that is imposed by degradation of RNA.These results are limited to the Agilent 44 k microarray platform and should be carefully interpreted when transferring to other settings.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department Medical Statistics, University Medicine Göttingen, Humboldtallee 32, 37073 Göttingen, Germany.

ABSTRACT

Background: Gene expression profiling is a highly sensitive technique which is used for profiling tumor samples for medical prognosis. RNA quality and degradation influence the analysis results of gene expression profiles. The impact of this influence on the profiles and its medical impact is not fully understood. As patient samples are very valuable for clinical studies, it is necessary to establish criteria for the RNA quality to be able to use these samples in later analysis.

Methods: To investigate the effects of RNA integrity on gene expression profiling, whole genome expression arrays were used. We used tumor biopsies from patients diagnosed with locally advanced rectal cancer. To simulate degradation, the isolated total RNA of all patients was subjected to heat-induced degradation in a time-dependent manner. Expression profiling was then performed and data were analyzed bioinformatically to assess the differences.

Results: The differences introduced by RNA degradation were largely outweighed by the biological differences between the patients. Only a relatively small number of probes (275 out of 41,000) show a significant effect due to degradation. The genes that show the strongest effect due to RNA degradation were, especially, those with short mRNAs and probe positions near the 5' end.

Conclusions: Degraded RNA from tumor samples (RIN > 5) can still be used to perform gene expression analysis. A much higher biological variance between patients is observed compared to the effect that is imposed by degradation of RNA. Nevertheless there are genes, very short ones and those with the probe binding side close to the 5' end that should be excluded from gene expression analysis when working with degraded RNA. These results are limited to the Agilent 44 k microarray platform and should be carefully interpreted when transferring to other settings.

Show MeSH
Related in: MedlinePlus