Limits...
Reference gene validation for RT-qPCR, a note on different available software packages.

De Spiegelaere W, Dern-Wieloch J, Weigel R, Schumacher V, Schorle H, Nettersheim D, Bergmann M, Brehm R, Kliesch S, Vandekerckhove L, Fink C - PLoS ONE (2015)

Bottom Line: Similar results were obtained by the three approaches for the most and least stably expressed genes.Interestingly, different outputs were obtained between the original software packages and the RefFinder tool, which is based on raw Cq values for input.New software tools that incorporate these algorithms should be carefully validated prior to use.

View Article: PubMed Central - PubMed

Affiliation: Ghent University, Department of Internal Medicine, Ghent, Belgium.

ABSTRACT

Background: An appropriate normalization strategy is crucial for data analysis from real time reverse transcription polymerase chain reactions (RT-qPCR). It is widely supported to identify and validate stable reference genes, since no single biological gene is stably expressed between cell types or within cells under different conditions. Different algorithms exist to validate optimal reference genes for normalization. Applying human cells, we here compare the three main methods to the online available RefFinder tool that integrates these algorithms along with R-based software packages which include the NormFinder and GeNorm algorithms.

Results: 14 candidate reference genes were assessed by RT-qPCR in two sample sets, i.e. a set of samples of human testicular tissue containing carcinoma in situ (CIS), and a set of samples from the human adult Sertoli cell line (FS1) either cultured alone or in co-culture with the seminoma like cell line (TCam-2) or with equine bone marrow derived mesenchymal stem cells (eBM-MSC). Expression stabilities of the reference genes were evaluated using geNorm, NormFinder, and BestKeeper. Similar results were obtained by the three approaches for the most and least stably expressed genes. The R-based packages NormqPCR, SLqPCR and the NormFinder for R script gave identical gene rankings. Interestingly, different outputs were obtained between the original software packages and the RefFinder tool, which is based on raw Cq values for input. When the raw data were reanalysed assuming 100% efficiency for all genes, then the outputs of the original software packages were similar to the RefFinder software, indicating that RefFinder outputs may be biased because PCR efficiencies are not taken into account.

Conclusions: This report shows that assay efficiency is an important parameter for reference gene validation. New software tools that incorporate these algorithms should be carefully validated prior to use.

No MeSH data available.


Related in: MedlinePlus

geNorm outputs with efficiency corrected data (A&C) and without efficiency corrected data (B&D) for the two datasets, i.e. FS1 (A&B) and CIS (C&D).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4380439&req=5

pone.0122515.g001: geNorm outputs with efficiency corrected data (A&C) and without efficiency corrected data (B&D) for the two datasets, i.e. FS1 (A&B) and CIS (C&D).

Mentions: To assess the influence of using efficiency corrected versus non-corrected values in the validation of reference genes, a comparison of the rankings was made between efficiency corrected and non-corrected relative quantities in sets of samples (Table 2). This comparison revealed that 71% (20/28) and 50% (14/28) of the rankings were different using geNorm (Table 2, Fig 1) and NormFinder (Table 2, Fig 2), respectively. A classification based on minor changes (position switches of one rank up or down) or major changes (position switch of more than one rank up or down) revealed that 16 out of 20 discrepant rankings were major changes with the geNorm algorithm, whereas only 5 out of 14 position changes were major with the NormFinder algorithm. In the FS1 data set the three most stable reference genes as appointed by the geNorm algorithm, SDHA, HMBS and UBC, were replaced by TOP2B, B2M and HPRT1 when gene specific efficiency was not taken into account. Similarly, in the CIS samples, YWHAZ, TOP2B and HMBS would be replaced by ACTB, HMBS and SDHA. The effect on the top three most stable genes with the NormFinder algorithm was less affected by the exclusion of the gene specific efficiency. In the FS1 samples the original top three, HMBS, TBP and UBC, would change to HMBS, TBP and HPRT1 and in the CIS samples the initial top three, HMBS, ACTB and YWHAZ would change to HMBS, IGF1R and ACTB.


Reference gene validation for RT-qPCR, a note on different available software packages.

De Spiegelaere W, Dern-Wieloch J, Weigel R, Schumacher V, Schorle H, Nettersheim D, Bergmann M, Brehm R, Kliesch S, Vandekerckhove L, Fink C - PLoS ONE (2015)

geNorm outputs with efficiency corrected data (A&C) and without efficiency corrected data (B&D) for the two datasets, i.e. FS1 (A&B) and CIS (C&D).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4380439&req=5

pone.0122515.g001: geNorm outputs with efficiency corrected data (A&C) and without efficiency corrected data (B&D) for the two datasets, i.e. FS1 (A&B) and CIS (C&D).
Mentions: To assess the influence of using efficiency corrected versus non-corrected values in the validation of reference genes, a comparison of the rankings was made between efficiency corrected and non-corrected relative quantities in sets of samples (Table 2). This comparison revealed that 71% (20/28) and 50% (14/28) of the rankings were different using geNorm (Table 2, Fig 1) and NormFinder (Table 2, Fig 2), respectively. A classification based on minor changes (position switches of one rank up or down) or major changes (position switch of more than one rank up or down) revealed that 16 out of 20 discrepant rankings were major changes with the geNorm algorithm, whereas only 5 out of 14 position changes were major with the NormFinder algorithm. In the FS1 data set the three most stable reference genes as appointed by the geNorm algorithm, SDHA, HMBS and UBC, were replaced by TOP2B, B2M and HPRT1 when gene specific efficiency was not taken into account. Similarly, in the CIS samples, YWHAZ, TOP2B and HMBS would be replaced by ACTB, HMBS and SDHA. The effect on the top three most stable genes with the NormFinder algorithm was less affected by the exclusion of the gene specific efficiency. In the FS1 samples the original top three, HMBS, TBP and UBC, would change to HMBS, TBP and HPRT1 and in the CIS samples the initial top three, HMBS, ACTB and YWHAZ would change to HMBS, IGF1R and ACTB.

Bottom Line: Similar results were obtained by the three approaches for the most and least stably expressed genes.Interestingly, different outputs were obtained between the original software packages and the RefFinder tool, which is based on raw Cq values for input.New software tools that incorporate these algorithms should be carefully validated prior to use.

View Article: PubMed Central - PubMed

Affiliation: Ghent University, Department of Internal Medicine, Ghent, Belgium.

ABSTRACT

Background: An appropriate normalization strategy is crucial for data analysis from real time reverse transcription polymerase chain reactions (RT-qPCR). It is widely supported to identify and validate stable reference genes, since no single biological gene is stably expressed between cell types or within cells under different conditions. Different algorithms exist to validate optimal reference genes for normalization. Applying human cells, we here compare the three main methods to the online available RefFinder tool that integrates these algorithms along with R-based software packages which include the NormFinder and GeNorm algorithms.

Results: 14 candidate reference genes were assessed by RT-qPCR in two sample sets, i.e. a set of samples of human testicular tissue containing carcinoma in situ (CIS), and a set of samples from the human adult Sertoli cell line (FS1) either cultured alone or in co-culture with the seminoma like cell line (TCam-2) or with equine bone marrow derived mesenchymal stem cells (eBM-MSC). Expression stabilities of the reference genes were evaluated using geNorm, NormFinder, and BestKeeper. Similar results were obtained by the three approaches for the most and least stably expressed genes. The R-based packages NormqPCR, SLqPCR and the NormFinder for R script gave identical gene rankings. Interestingly, different outputs were obtained between the original software packages and the RefFinder tool, which is based on raw Cq values for input. When the raw data were reanalysed assuming 100% efficiency for all genes, then the outputs of the original software packages were similar to the RefFinder software, indicating that RefFinder outputs may be biased because PCR efficiencies are not taken into account.

Conclusions: This report shows that assay efficiency is an important parameter for reference gene validation. New software tools that incorporate these algorithms should be carefully validated prior to use.

No MeSH data available.


Related in: MedlinePlus