There been no question about deta quality and reliability with NGS, which are still there with miRNAs. One positive side is, with appropriate strategy one can reliably go for novel RNA findings which is not possible with microarray. Again with strategy like SAGE one can acually get exact count of tags and thus much robust expression value, miCross cant be any match to this...

Microarray@

I totally agree that RNAseq can help identify novel transcripts, and also identify gene fusion and mutations that is not possible for microarray. But what I'm asking is specifically in the pooled shRNA screening scenario, where the only thing we are measuring is the relative abundance of the sequence tags designed and attached to the shRNAs. For this purpose, except better data quality I didn't see any unique value to use NGS rather than microarray.

with my limited experience,I guess what you just told is is true

Depending on your assay the benefits can include quality of data and flexibility (ability to use different/additional shRNA with redesigning a chip or to use focused sets without wasting space on a genome chip). Also, some of the complexity of data analysis is being addressed as people have created software specifically for pooled screens.

There was a really nice publication Genome Biology last year titled "High-throughput RNA interference screening using pooled shRNA libraries and next generation sequencing." It discusses the benefits of NGS including sensitivity, dynamic range and representation of the library. There is a detailed walk through of their screening process, and they compare their data to what may expected from an array.

They also developed an open source program called shALIGN. I have not used it, but they describe some of the data analysis problems it solves because it was specifically designed to align sequences to shRNA libraries rather than aligning to the whole genome.

You also gain flexibility in the pool content. You can add new shRNAs to your screen, change libraries used to create the pools, or do more focused screens on gene families and pathways without having to print new arrays or deal with wasted space on the array. Smaller, more focused screens may also help with the data analysis.

Both methods are clearly valid and you may benefit more or less depending on what array your are comparing too. Another big factor in any pooled screen is the quality of the shRNA library. If only a fraction of the shRNAs are functional you may benefit more from a more sensitive readout since the nonfunctional hairpins decrease your effective representation and add to the noise.

The major advantage of using shRNA sequencing is the wide dynamic detection range as compared to microarray. But as always, it will increase the sensitivity as well as the noise. To me the major hurdle in shRNA screening is not the end point detection method but rather the PCR bias caused by the hairpin structure. That's why nowaday barcodes have been introduced to shRNA library so that one can avoid the hairpin regions during PCR before sequencing.

I agree with Mengchu that PCR bias is a big concern and using barcode rather than amplifying the hairpin may help to reduce this bias. If ignore the better data quality - wider dynamic range, linearity, etc., this bias, because of hairpin or because of barcode, will affect both microarray and NGS in the same way. So I think flexibility, as Andy discussed above, maybe the biggest advantage of NGS over microarray in pooled shRNA screenings.