Print |

You have not viewed any products recently.

 

FAQs

FAQ Categories

Experiment Design

Sample Submission

LncRNA Research

circRNA Research

Epigenetic Research

miRStar™ PCR System

Data Analysis

Questions & Answers

Experiment Design

Why do I need biological replicates? Isn’t one sample per group enough?

Biological replicates increase the probability that your array results are statistically significant. As with any experiment in biology, genetic differences exist between individuals with the same apparent phenotype, even in inbred laboratory strains. If you were to test only one individual per experimental group, you would not be able to conclude with confidence that any observed difference is due to the difference in experimental conditions. It is always possible that an underlying, otherwise “invisible” genotypic variation is contributing to your results. Only by examining more than one individual per group can you determine whether or not a differential expression result is real. This is especially important in order to reduce the occurrence of false negatives, which you will never see.

back to top ↑

How many biological replicates per group do you recommend?

We strongly recommend a minimum of three biological replicates per group to achieve the most statistically significant data. There are two major reasons for why we do not recommend using fewer than three. First, if you had only 1 sample per group, there is a chance, that that one individual represents an “outlier”. However, since you only tested one sample in that group, you would never know whether it is an outlier or not. Further, if you had two biological replicates per group, one of those individuals could represent the outlier. However, if such was the case, you would not know which of the two individuals in that group represented the outlier. Only if you have three or more replicates per group can you be reasonably confident that any genetic differences observed between groups are real, because it is highly unlikely to have two outliers in a group of three. For mammalian tissues, we prefer six biological replicates per group, due to the heterogeneity of cell populations in tissues.

back to top ↑

Do you need to run technical replicates?

No. The arrays have been validated many times, and contain control probes. So, technical replicates are a waste of money.

back to top ↑

Sample Submission

How to determine whether or not RNA extracted from FFPE samples is suitable for expression profiling?

Due to the severe degradation of RNA extracted from FFPE samples, it is unlikely that one can obtain an accurate RIN from the Agilent Bioanalyzer. Instead, we run the RNA on a MOPS-formaldehyde gel and visually inspect the ribosomal RNA bands. If we are able to see even traces of these ribosomal RNA bands on the gel, we go ahead and proceed with the expression profiling experiment.

back to top ↑

For FFPE samples, should we send FFPE samples or the extracted RNA?

The main problem with FFPE samples is that the RNA is typically degraded. Therefore, obtaining the highest yield and quality of RNA from the de-paraffinization and extraction procedures is essential for obtaining high quality profiling data, for both microarrays and sequencing. Arraystar has many years of experience performing extraction of RNA from FFPE samples. Therefore, we strongly recommend that you send us your FFPE samples, and we will extract the RNA for you using our optimized protocols.

back to top ↑

Which method do Arraystar recommend to extracting total RNA from samples?

At Arraystar, we use TRIzol® Reagent from Life Technologies for extraction of total RNA from cells and tissues (http://www.lifetechnologies.com/order/catalog/product/15596026). However, if you have experience with total RNA isolation, then we recommend using whichever method you are most comfortable with. There are several other good RNA extraction methods as well, such as Qiagen’s RNeasy kit (http://www.qiagen.com/qdm/rna/rneasy-plus-kits?cmpid=Qven10GARneasy). The most important objective for total RNA isolation is to obtain as high a yield as possible with minimal degradation and contamination from proteins, DNA, and organic solvents, regardless of the extraction method used.

back to top ↑

Should I submit total RNA or isolated miRNA for Arraystar’s miRNA seq and miRNA PCR array services?

  1. Our seq and miRStarTM systems have been specifically optimized for use on total RNA extracted using Trizol method.
  2. To avoid miRNA loss during purification, we recommend you submit total RNA not small RNAs.
  3. If you have isolated miRNA, technically we can do it, but we shall not assure the success of sequencing library preparation that is affected by the extraction efficiency of small RNA. We highly recommend that you should monitor the extraction efficiency with Agilent 2100 Bioanalyzer, and provide the results to us.

back to top ↑

LncRNA Research

Is too provocative to think that we may discover coding function for the lnc RNA after all?

It is still a challenge to distinguish LncRNAs from mRNAs, and there is currently no “perfect” tools or databases available for the reliable collection of LncRNAs. To solve this problem, Arraystar has developed a stringent computational pipeline to reliably identify lncRNAs. All the candidate transcripts were carefully filtered by known annotations and positive coding potential, and any candidate containing a substantial open reading frame (ORF) covering 35% or more of its length and containing Pfam/Tigfam protein domains is rejected.

back to top ↑

Is there any database available to compare the expression of lncRNAs between tumor tissues and match normal tissues?

Unfortunately, there is no such kind of database available at present.

back to top ↑

Does each lncRNA have a polyA tail?

Most of the lncRNAs (75%) have a poly-A tail, while some of them don't.

back to top ↑

Do all LncRNAs must have biological function?

We cannot make an assertion that all LncRNAs must have biological functions. Although the functional roles of LncRNAs are largely unknown, many LncRNAs have been discovered having important biological functions. A significant number of them have also been shown to exhibit cell type-specific expression, precise localization to subcellular compartments, and deregulation in various human diseases, suggesting that the majority of LncRNAs are likely to be functional. Some could be used as biomarkers in lieu of a precise biological function.

back to top ↑

What databases does Arraystar use as the sources of LncRNAs on the LncRNA microarrays?

Arraystar uses NCBI RefSeq, UCSC known genes, RNAdb2.0, NRED, UCRs, GENCODE, and FANTOM (mouse) as the source databases for LncRNAs. Additional LncRNAs, primarily the lincRNAs (long intergenic noncoding RNAs) are taken from landmark publications. After database selection, highly similar sequences and non-coding RNAs shorter than 200 bp are excluded. For the protein-coding genes, we use NCBI RefSeq and CCDS. CCDS is a collaboration between NCBI, the European Bioinformatics Institute, the Wellcome Trust Sanger Institute, and the University of California, Santa Cruz.

back to top ↑

What are the positive control probes on the human LncRNA microarray?

We designed probes to detect 20 housekeeping genes as positive controls, and are replicated 10 times. These 20 genes are NM_002455, NM_021009, NM_006013, NM_001536, NM_003746, NM_002107, NM_000754, NM_003753, NM_001003, NM_001101, NM_006098, NM_002539, NM_022551, NM_004309, NM_002046, NM_001861, NM_000291, NM_005022, NM_001614 and NM_000841.

back to top ↑

What are the negative control probes on the human LncRNA microarray?

We designed negative control probes, which are named “NegativeXXX”.

back to top ↑

For the coding mRNA on the array, is it just the exon regions or is it the whole gene locus (introns and exons)?

For each protein-coding mRNA, we designed a probe targeting their specific exon or exon-junction regions.

back to top ↑

How does the microarray identify the different isoforms? Are there different wells for each probe?

Each array probe is designed to hybridize to a region unique to the isoform or crossing the exon splice junction. Each transcript isoform from the same gene has its own unique array probe. These probes are individually located on the array.

back to top ↑

Which should I use for Long non-coding RNA (LncRNA) profiling: Microarray or Next Generation Sequencing?

There are several scientifically-sound reasons for why you should choose microarrays over next-generation sequencing for LncRNA expression profiling. First, it has been shown that most LncRNAs are expressed at significantly lower levels than mRNAs (1, 2). Therefore, many more sequencing reads are required for LncRNA analysis than for mRNAs due to their low abundance. Recent publications used more than 120 million raw reads per sample in order to obtain acceptable coverage of LncRNAs (2,3). Second, only 1,000-4,000 LncRNAs are detected by greater than 120 million sequencing reads (3), while 7,000-12,000 LncRNAs are normally detectable using Arraystar LncRNA microarrays. Therefore, LncRNA microarrays can detect more LncRNAs than Next Generation Sequencing, at a lower cost.

References

  1. Guttman, et al. (2010). Nature Biotechnology 28, 503.
  2. Cabili, et al. (2011). Genes Dev. 25, 1950.
  3. Markus et al. (2012). Genes Dev. 26, 338-343

back to top ↑

How can I obtain the sequences of the LncRNAs I am interested in studying further?

Each of the LncRNAs represented on the Arraystar LncRNA microarrays is annotated in one of several different public databases. Each of these databases has its own unique procedure for retrieving sequence information from a given gene. Arraystar provides a separate document that gives instructions on how to obtain the sequences of LncRNAs of interest. Please contact us.

back to top ↑

Are there any integrated database similar to GeneCards, Biocarta, Kegg to link the LncRNAs with cancer or other pathways? Existing LncRNA data mining tools are not very user friendly?

There are several LncRNA databases or online resources such as lncRNAdb (http://www.lncrnadb.org/) or LncRNADisease (http://202.38.126.151/hmdd/html/tools/lncrnadisease.html) that offer some of the functionalities.

back to top ↑

circRNA Research

Could you send us at least the annotations of all circular RNAS probesets that are expressed in our samples after filtering?

The annotations for the pass-filtering circRNAs have already been supplied in the "CircRNA Expression Profiling Data.xls" file in the "File 2. Data" folder.

back to top ↑

Are there any validation recommendations for circular RNA expression analysis?

  1. Design outward-facing primers
  2. Perform reverse transcription with random hexamer primer.
  3. Treat total RNA with RNase R to remove linear RNAs, thus reducing false positive signals.

You may refer to "Circular RNAs are abundant, conserved, and associated with ALU repeats" for detailed information.

back to top ↑

Does Arraystar provide the complete annotation of array probe sets (at least genomic coordinates and gene names) that are needed to compare the Arraystar results with our RNA seq data?

 Complete annotations for all circRNAs are not disclosed. However, the probe IDs and the probe sequences, but not annotation for the genes, will be deposited as a GEO platform in the NCBI GEO repository.

back to top ↑

Does Arraystar provide circular RNA expression vector?

We do not supply individual circRNAs.  However, investigagtors may construct, transfect and produce recombinant circular RNA using circular RNA expression vector by themselves. A description of recombinant circular RNA production can be found in the publication:

Liang and Wilusz (2014) "Short intronic repeat sequences facilitate circular RNA production" Genes Dev. 2014 Oct 15; 28(20): 2233–2247. [PMCID: PMC4201285]

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4201285/

back to top ↑

Could you please send me the names of the circRNAs corresponding to each probe in the array?

The circular RNAs on the array were collected from various sources. The "CircRNA Expression Profiling Data.xls" contains circRNAs expressed in your samples. You will find "probeID" and the Arraystar consolidated "circRNA" names (Column T). The same information is also available in "Differentially Expressed CircRNAs.xls" for the differentially expressed circRNAs. Additionally, the sequences for differentially expressed circRNAs are provided in "Differentially expressed circRNA sequences.fa". Also, if the "source" column is "circBase", you can use the "alias" as the accession ID to search circBase.org records. For circRNAs never expressed in your samples, the annotation is limited to the probeIDs as in the raw data.

back to top ↑

Epigenetic Research

What are the quality control steps to determine whether the hMeDIP procedure was successful? Are there any particular genes Arraystar tests for with qPCR that are constitutively hydroxymethylated in normal tissue irrespective of origin?

Before immunoprecipitation, a synthetic oligonucleotide containing 5hmC at CpG positions is mixed with the fragmented genomic DNA as a spike-in positive control. The enrichment of hydroxymethylated DNA by immunoprecipitation is assessed by qPCR using the positive control sequence.

back to top ↑

In the event that hMeDIP capture fails for any given submitted sample or in the event that sequencing or array analysis is performed, but peaks/hits are of low quantity and quality, what is the procedure that Arraystar would follow?

 If the hMeDIP-qPCR shows poor IP efficiency, we will repeat the IP experiment till the IP efficiency is achieved. However, we do not guarantee the peak quantity, because hydroxymethylation levels vary largely across different sample types.

back to top ↑

How to ensure the reliability of MeDIP experiment?

Several stringent quality control measures are implemented:

  • Sample DNA QC to ensure the submitted samples meet the amount and quality requirements of MeDIP.
  • Sonication QC to ensure the genomic DNA is fragmented to the desired size distribution.
  • H19 imprinting gene is used as the positive control of DNA methylation.
  • GAPDH housekeeping gene is used as a negative control of DNA methylation.
  • qPCR QC on the MeDIP and input DNAs of the positive and negative control gene loci to ensure the enrichment efficiency and specificity.
  • QC steps for microarray (MeDIP-chip) or sequencing (MeDIP-seq) from the MeDIP materials.

back to top ↑

miRStar™ PCR System

Could you please describe me briefly the miRStar™ PCR system?

In miRStar™ system, an adapter is ligated to the 3' end of miRNAs, and then, RT-PCR is performed by using the miRNA-specific Forward Primer and Universal Reverse Primer (Figure 1).  The original miRNA-specific Forward Primer is designed according to the complementary sequence of the corresponding mature miRNA. Then, it is suitably truncated if the original Tm is > 60℃ or is appropriately lengthened based on the known adaptor sequence if the original Tm is <60℃ (Figure 1).  All the optimized miRNA-specific Forward Primers have been experimentally validated to ensure their high specificity, assay uniformity, and that the Tm values are ~60℃(Figure 3).

 

back to top ↑

What´s the Tm for the miRStar™ PCR panel

The Tm values of all the primers on miRStar™ PCR panel have been optimized to 60℃.

back to top ↑

How long are the primers?

The optimized primers are 19-25 nt in length.

back to top ↑

Data Analysis

How are the fold changes in a microarray experiment calculated? In my Excel file neither the raw nor the normalized signal intensities seem to correlate with the Fold change values.

 First, we do not use raw signal intensity values to determine fold changes. We only use the normalized values. Second, those normalized intensity values are log2 transformed. We calculate Fold Change by subtracting the normalized intensity value of one group from the normalized intensity value of another group. For example, if the normalized intensity of the control group is 3.5819511 and the normalized intensity of the experimental group is 5.984653. Subtracting the normalized intensity of the control group from the normalized intensity of the experimental group gives you 5.984653 - 3.5819511 = 2.4027019. Since the normalized intensities are actually log2 transformed, this result is actually the log2 of the fold change. Thus, the actual fold change is 2^2.4027019=5.2879257, which is the level of fold increase of expression in the control group vs. the experimental group. This is the value indicated in the fold change column labeled “FC (abs)”.

back to top ↑

In the Table entitled “Differentially expressed LncRNAs /mRNA.xls”, there are many entries (rows) are repeating and each counts an individual entry. I do not know how such things happen and how meaningful the data in this table would be.

  1. Arraystar divides all known LncRNAs into subgroups, based on their genomic organization with nearby protein-coding genes. These are: Sense-overlapping, antisense-overlapping, bidirectional, intronic, and intergenic. Some LncRNAs are closely associated with more than one transcript variant of the same coding gene. We listed all the transcripts of the gene in Column Q for your reference.
  2. Occasionally we see an LncRNA that is physically associated with more than one protein-coding gene, and thus might occur in more than one subgroup, or in the same subgroup multiple times.

back to top ↑

If control groups and experimental groups were hybridized on slides of different batch, how do you control batch effets?

The arrays were produced by Agilent manufacturing process and under stringent quality control, which guarantee very low technical variations among different arrays in the same production batch. Arraystar performs every single project using the same batch of arrays, reagents, kits, procedures and lab personnel, to minimize any batch effects. Biological variations far exceed any array technical variations. More importantly, array data across the arrays are quantile normalized for comparison; they are not compared by raw intensities. Any between-array technical difference has no effect on comparison. In other words, you do not need to concern about how the samples are arranged on the arrays.

back to top ↑

Is there a way to identify individual transcripts from the volcano plot, in other words, have a link from the plot to the Excel file?

It is possible to identify transcripts from the volcano plot. However, this can only be done with the GeneSpring analysis software.

back to top ↑

What scanner is Arraystar using for scanning the hybridization images?

We used the Agilent scanner model G2505C.

back to top ↑

What software is Arraystar using to extract raw and normalized data from microarrays?

For acquisition of array images, we use Agilent Feature Extraction software version 11.0.1.1. For data analysis, including data normalization and differential expression, we use Agilent GeneSpring GX.

back to top ↑

How are sample intensities on a microarray normalized?

The raw signal intensities of all samples are normalized using one of two methods, which are carried out in Agilent’s GeneSpring software package. The first method is “Median normalization”, in which the software divides the signal intensity of each gene by the median intensity of all the genes on the chip. The second method is “Quantile normalization”, which is used when there are a significant number of outliers. Most of our project data is Quantile-normalized.

back to top ↑

How do you determine/rank the changes?

When we perform a microarray experiment, the arrays are scanned by a scanning machine, which assigns a raw intensity value for each transcript represented on the array. Next, using "flags", we discard any transcript that does not appear to be expressed in any of your samples (either low signal/noise ratio or lack of signal and/or background uniformity). We then take the remaining raw signals and normalize them to the median intensity of all the signals on the array (or, alternatively, by quantile normalization). Finally, we use those normalized signal intensities for the fold change comparisons between groups. As mentioned above, we will set a fold change cutoff value, which is typically 2.0, but that can be adjusted if desired. We also set a cutoff of 0.05 for the p-value, but that too can be adjusted.

We typically don't rank the hits from highest to lowest or vice versa, but that is not an issue here. We provide the data in Excel, so you can easily set a rank order using the software.

back to top ↑

How to pick differentially expressed transcripts to validate?

Researchers can take at least two different approaches to validating their microarray data, depending on their individual research needs. One way is to validate all of the differentially expressed transcripts on the array, in an unbiased, high-throughput manner. Another way is to just focus on those differentially expressed transcripts that appear to be associated with a biological function of interest, using the GO and Pathway analysis data provided in your report. However, this latter approach might not be very effective in the case of LncRNAs, because there are a great many LncRNAs whose function and roles in cellular processes are unknown. In any case, we strongly recommend eliminating those transcripts that have extremely low raw signal intensities, and then picking transcripts with larger fold change, as well as lower p values.

back to top ↑

Why is one probe listed multiple times on my LncRNA microarray results?

Some LncRNAs are closely associated with more than one protein-coding gene.  Such LncRNAs are listed more than once in order to show the annotation information for all the associated protein-coding genes.

back to top ↑

I noticed that a gene I was expecting to see expression data for was not present in my report. Does it have a different name in the report, or is it not represented on the microarray?

One reason a given gene appears to be “not present” on the array is that its probe signal intensity did not meet our “flagging” criteria to be included in the analysis. The software we use to analyze the data assigns each probe in a particular sample one of three flags: “Present” (P), “Marginal” (M), or “Absent” (A). Probes are flagged as “A” based on one or more of the following criteria: 1) the signal intensity is too low; 2) the signal intensity is not uniform; 3) the background is too high; or 4) the background is not uniform. Preliminarily, “A” flagged probes are considered to be poorly or not expressed in that sample, while those flagged either “P” or “M” are considered to be expressed at a significant level. When we indicate the signal intensities in the Excel files in your report, we tell you the cutoff that was used in order for each probe to be included in the analysis. For example, if the cutoff for including probes in the analysis we use requires that at least 6 out of 9 samples have flags of “P” or “M”, and a given gene has been flagged as “A” in 5 samples, then it will not be included in the analysis, even though it is represented on the microarray.

back to top ↑