Reader Comments

Post a new comment on this article

Clarifications and data upload requested

Posted by kwitwer1 on 06 Apr 2012 at 20:09 GMT

After performing microarray profiling of RNA from cells exposed or not to a chimeric murine virus known as XMRV, Mohan, et al. suggest that microarray analysis "clearly demonstrates that microRNAs are regulated during XMRV infection," specifically miR-193a-3p and a putative miRNA called E1245. However, the lack of biological or technical replicates, statistical analysis, and validation assays might make it difficult for some to accept this conclusion.
Presenting PCA in Figure 2, the authors state that "the overall miRNA expression profile in LNCaP and DU145 cell lines form a tight cluster suggesting minimal variation due to time and infection status of these two cell types, while the PBLs and MDM profiles are more spread, indicating the global miR levels being affected both due to time and infection status." The final assertion seems to be somewhat inconsistent with Figure 2, showing that infected and uninfected PBL and MDM generally pair up based on time point, not infection status. A slightly different conclusion might be that cancer cell lines, being relatively clonal, tend to display less miRNA variation during culture than heterogeneous populations of primary blood cells, and that the presence of the chimeric murine virus has no readily apparent effect. This is also clear from the heatmap in Figure 3. Technical processing differences between the time points could be an equally valid explanation for apparent time point-associated expression differences.
Several clarifications would be much appreciated and would help me (and perhaps other readers) to understand the study better:
1. Because no statistical evidence of differential expression was found, the authors determined the difference between mock and infected conditions for each group of cells at the 6, 24, and 48 hour time points and report a single, undefined "abs" value in the Figure 5 table. What is the "abs" value? Is it the average difference between log(2) values for infected and uninfected samples for all time points? If so, it would be useful to know the direction of regulation as well as how much this difference varied between the time points. Differences (at 6, 24, and 48 hours) of 2, 2.2, and 2.1 average 2.1, but so do 0.1, 6.1, and 0.1. Some might consider the former case to be more convincing than the latter. A small amount of additional information could greatly strengthen the authors' conclusions.
2. In Figure 3, it is not clear what the authors mean by including "all microRNAs with standard deviation over 1". Further clarification would be useful. Also, although miR-193a-3p is depicted, where is the proprietary putative miRNA "E1245" in the heatmap?
3. The utility of microarray studies to the scientific community is greatly enhanced by availability of the data; hence, most journals, including PLoS ONE, mandate data submission to public databases such as GEO or ArrayExpress. The authors are strongly encouraged to upload their raw and processed array data as per journal policy (see and .)

No competing interests declared.

RE: Clarifications and data upload requested

CDAtreya replied to kwitwer1 on 23 Apr 2012 at 14:40 GMT

Response from authors

1. Inference from Principal Component Analysis: The authors agree that the cancer cell lines display less microRNA variation with regard to time and infection status and the manuscript does indicate this inference as well (page 3, lines 15-18 in the publication). Indeed, the miR profile-based PCA plot for PBLs and MDMs does indicate a more spread out profile with regard to time due to the heterogeneous nature of these cells and relatively tighter in the other two cell lines, since they are clonal in nature as pointed out by one of the readers. Nonetheless, it is also true that at each time point, the infected cells miR profile is moderately separated from the uninfected cells.

2. Yes, the Abs value is indicative of the average difference between Log2 ratios between infected and uninfected (Hy3/Hy5) samples for all time points. The direction of regulation is variable with subtle differences observed in both directions (up and down).

3. Figure 3: This heat map was generated by including the top 100 microRNAs with a SD value of >1.0 in order to capture significant variation between the four cell types for each microRNA. As correctly pointed out by one of the readers, the miRPlus-E1245 is not included in this heat map due to its lower SD value of 0.62, i.e. below 1 indicating a subtle but observable differential expression common to all 4 cell types (as demonstrated in Fig 6).

No competing interests declared.