PMC:3807239 / 4082-12364 JSONTXT

{"project":"2_test","denotations":[{"id":"24163754-23014960-112081888","span":{"begin":7678,"end":7680},"obj":"23014960"},{"id":"24163754-23014960-112081888","span":{"begin":7678,"end":7680},"obj":"23014960"},{"id":"24163754-11382363-112081889","span":{"begin":7761,"end":7763},"obj":"11382363"},{"id":"24163754-11382363-112081889","span":{"begin":7761,"end":7763},"obj":"11382363"},{"id":"24163754-11747612-112081889","span":{"begin":7761,"end":7763},"obj":"11747612"},{"id":"24163754-11747612-112081889","span":{"begin":7761,"end":7763},"obj":"11747612"},{"id":"24163754-12169537-69479651","span":{"begin":8036,"end":8038},"obj":"12169537"},{"id":"24163754-12169537-69479651","span":{"begin":8036,"end":8038},"obj":"12169537"},{"id":"24163754-23014960-69479648","span":{"begin":7678,"end":7680},"obj":"23014960"},{"id":"24163754-23014960-69479648","span":{"begin":7678,"end":7680},"obj":"23014960"},{"id":"24163754-11382363-69479649","span":{"begin":7764,"end":7766},"obj":"11382363"},{"id":"24163754-11382363-69479649","span":{"begin":7764,"end":7766},"obj":"11382363"},{"id":"24163754-11747612-69479650","span":{"begin":7767,"end":7769},"obj":"11747612"},{"id":"24163754-11747612-69479650","span":{"begin":7767,"end":7769},"obj":"11747612"},{"id":"T9251","span":{"begin":8036,"end":8038},"obj":"12169537"},{"id":"T23084","span":{"begin":8036,"end":8038},"obj":"12169537"}],"text":"2. Measurement Quality and Background Correction\n\n2.1. MiRCURY LNA MicroRNA Array\nTwo versions of miRCURY LNA microRNA arrays are used for miRNA profiling (see the Materials and Methods section for more details). All slides are scanned using an Axon Gene Pix Professional 4200A microarray scanner (Molecular Devices, Sunnyvale, CA, USA), and the images are gridded and analyzed using ImaGene 7.0 software (BioDiscovery Inc., Hawthorne, CA, USA). MiRCURY LNA microRNA Array v7.5.0 (LNAv7 hereafter) is used to profile 359 miRNAs for two HCT-116 cell lines. MiRCURY LNA microRNA Array v9.2 (LNAv9 hereafter) is used to profile 577 miRNAs for 10 osterosarcoma xenograft specimens. On each slide of the LNA arrays, there are four technical replicates for each miRNA. The background signals are estimated by measuring the intensity of the surrounding area (pixels in the local background region) of the corresponding spot masks, and the signal for an miRNA from a specific spot is approximated by the intensity measure from the local signal region. The expression of an miRNA is computed based on the (local) background subtracted signals from the four replicates. All spots are automatically flagged by the image processing software to check the signal quality.\nTable 1 shows the five-number summaries of the automatic flags for the spots (probes) for the miRNAs being tested based on thirteen (13) LNAv7 slides and forty-eight (48) LNAv9 slides, respectively. From Table 1, we find the following: Among the 13 LNAv7 arrays, on average, approximately 58% of the probes have reasonably strong signals (not flagged). In the worst case, about 42% of spots are not flagged, and approximately 57% of the spots are low-expressed or missing spots. One slide contains more than 77% of spots with no flags.Among the 48 LNAv9 arrays, on average, less than 20% of the probes have reasonably strong signals, while more than 50% of the probes are empty spots. The best slides have approximately 45% of non-flagged spots, while the non-flagged spots have less than 4% in the worst slide.For both LNAv7 and LNAv9, the proportions of poor spots (background/signal contaminated, high ignored percentage and others) are relatively low.\nmicroarrays-02-00034-t001_Table 1 Table 1 Quality flags with miRCURY LNA arrays (all hsa-miR probes). In terms of the flagged spots, it looks like LNAv7 arrays have better signal quality than LNAv9 arrays. One potential explanation for the observation is that more weakly or not expressed miRNAs are included in the LNAv9 arrays. To have a closer comparison, we illustrate the summary information for the flags of the 224 miRNAs tested by both LNAv7 and LNAv9 in Table 2. From Table 2, we find that even when comparing the same set of miRNAs among both versions, there was reduced signal in LNAv9 arrays. This difference could also be due to the use of different samples in the experiments: HCT-116 cell lines for LNAv7 and human osterosarcoma xenograft specimens for LNAv9, respectively.\nmicroarrays-02-00034-t002_Table 2 Table 2 Quality flags with miRCURY LNA arrays (overlapped hsa-miR probes).\n\n2.2. FlexmiR MicroRNA Human Panel\nA total of 319 human miRNAs are profiled using the bead-based Luminex FlexmiR MicroRNA Human Panel (Luminex, Corp., Austin, TX, USA) for 40 treated and untreated osterosarcoma xenograft specimens. Due to the capacity of the pool, all miRNAs are divided into five groups and are tested using five different human pools separately. The intensities are captured with a Luminex-200 instrument. In addition, for each microsphere type being tested, a background control (water treated in the same manner as an RNA sample) is used to measure the background signal (median fluorescence intensity, MFI). The system does not flag results based on the signal quality. For the 40 profiles based on the treated and untreated samples, we manually flag the signals of different miRNAs as follows: we compute the standard deviation, s, of the background signals for each pool. If the signal of an miRNA after background subtraction is smaller than 2s. but not smaller than s, we flag the miRNA as weakly expressed; if the signal of an miRNA after background subtraction is smaller than s, we flag it as empty; otherwise, an miRNA is not flagged. Table 3 shows the percentages of miRNAs based on all 40 profiles. From Table 3, we find that about ∼60% or more miRNAs have reasonably strong intensity measures across all 40 arrays. In the worst case, the total percentage of weakly or not expressed miRNAs is about 40%. In summary, compared with the LNAv7 and LNAv9 arrays, the signal quality of the bead arrays is much improved, as expected. But on the other hand, we see that the percentage of weakly or not expressed miRNAs is high in arrays from all three platforms.\nmicroarrays-02-00034-t003_Table 3 Table 3 Quality flags with FlexmiR bead arrays.\n\n2.3. Signal-to-Noise Ratio\nWe also compute the signal-to-noise ratios (SNRs) based on the signals that are not flagged to compare the signal quality of the three platforms. The ratio of the background-subtracted signal to the local estimated background signal is used to approximate the SNR. Figure 1 shows the boxplots of the logarithms of the SNRs for the LNA arrays and the bead arrays. The two panels to the left in Figure 1 show the boxplots of the probe-level SNRs for human miRNAs from LNAv7 and LNAv9, respectively. The right panel shows the miRNA-level SNRs for all miRNAs from the bead arrays. We find that in terms of SNR, the signal quality of LNAv7 arrays and FlexmiR bead arrays are relatively better than that of LNAv9 arrays. The signal quality of the bead arrays is expected to be better. But the difference between LNAv7 and LNAv9 could mainly be contributed to the different specimens used and some other experimental factors as well.\nFigure 1 Signal-to-noise ratio comparisons.\n\n2.4. Background Correction\nLet x={x1,x2,…,xn} and y={y1,y2,…,yn} be the true expressions of the entire miRNA population being tested in two cell populations—a control sample, which is an untreated specimen or a specimen treated with saline, and a test sample, which is a specimen treated with a drug. Sometimes, the control sample can also be a specimen from a subject without disease, and the test sample is a specimen from a subject with a certain disease. In practice, both x and y are not observable. Instead, x and y are usually measured using microarrays or other analytical platforms, and the corresponding measurements. X and Y. are usually coupled with errors. It is common in practice that x and y are approximated by simply subtracting the local estimates of the background noises from the intensity measures from the corresponding spots. Or, if replicates exist, the median or mean values will be used for different miRNAs, respectively. For miRNA microarrays, background correction via local background subtraction may cause difficulties in data analysis. First, the majority of miRNAs are weakly or not expressed, and as a result, background-subtraction using local estimates may result in negative values in x and y, which is not acceptable. Second, ignoring the probes that are flagged as weakly expressed may result in too many missing values or over-estimate the true expression levels of some miRNAs—either way will introduce significant bias to the detection of differentially expressed miRNAs.\nIt is commonly observed that the standard deviation of the measurements in microarray rises proportionally to the expression level. However, for totally unexpressed genes, this proportionality won’t continue down to zero, due to the fact that measurement errors always exist [16]. Motivated by these observations, two measurement error models are introduced [17,18,19]. Both models describe the measurement error for a given transcript for gene expression arrays using two error components—the multiplicative and additive errors. A generalized logarithm transformation (GLOG) is proposed for gene-expression microarray data analyses [20]. The GLOG can stabilize the variance of the measurements and take care of the measures of the weakly expressed miRNAs quite well and, hence, are suitable for miRNA microarray data analysis (more details in the Materials and Methods section).\n"}