Comparing data across different chemistries/throughputs.

[sameet]

Hi All,

I have data from 6 RNA-Seq runs that we did sometime last year on a HiSeq machine. A month ago we tried the new MiSeq sequencer and I have one data-set from the new machine. How valid would be comparison between the two data-sets. Specifically, I have the following question.

Would it be valid to compare these two data-sets. Especially considering that we get FPKM which would technically make comparison possible as the data-sets are independent of read-depth.

Thanks in advance.

[westerman]

Hum. I haven't been on SeqAnswers for several days and am sort of surprised that no-one answered your question. I see no problems with comparing the two data sets assuming that the FPKMs are valid.

[JackieBadger]

in one of our studies we compared data from the 454 FLX and 454 Junior machines. There is no reason why results should not be comparable. Plenty of studies compare and use data from different chemistries..... You should however do a thorough assessment of the error distributions between them. If they are not comparable, that is interesting in its own right.
We are moving from 454 to MiSeq at the moment...one thing will be to compare the results between these data sets. These kind of questions need to be looked at with more rigour. Seeing as I am an Illumina greenhorn...are FPKM values some kind of quality score for contigs or sequences?
If so, you are right that you really want to be comparing data of similar quality. We omitted a bunch of data which despite being at the same quality as a replicate data set, the mean depths were significantly different thus making accurate comparisons impossible. This all comes down to careful consideration of what goes into a run.
j