Why Fluidigm C1 gives more read counts than iCell8 (wafergen)

Question

[Za]

Hi,

I have single cell RNA-seq data for 9 time points ;I have one time point sequenced by Fluidigm C1 and 8 time points by iCell8 (wafergen) . Just recently I noticed a big difference in read counts between time points as Fluidigm C1 has given me more read counts than iCELL8. I am not sure if iCell8 (wafergen) gives less read counts or how to explain that. anyone knows the reason behind this difference in number of genes detected?

[Kevin Blighe]

I have answered here: C: The interpretation of PCA

Well, they are different technologies, so, they are measuring expression on different scales, undoubtedly. You should definitely assume that there will be a batch effect between these data-types, and correct for it appropriately if you are planning to merge these togeter.

May I ask how on Earth an experimental design was devised that includes data from 2 different platforms?

[Za]

You alright but my boss urges at this combination :( :(

[h.mon]

You may have better luck asking this at SeqAnswers, as they have more users with expert knowledge in wet-lab technologies. You could also read the literature comparing single cell systems.

But bear in mind the difference may also be biological: maybe the "Fluidigm time point" represents a cell state with more transcription? You can't untangle technical and biological factors with your design.

[Kevin Blighe]

Ah... I thought so. Those are never easy situations, an they can become problematic!

The best that you can do is to include 'technology' (FLUIDIGM | iCELL) as a covariate when you are processing these data. For example, the design fomula would be:

design = ~ condition + technology

[WouterDeCoster]

I have one time point sequenced by Fluidigm C1 and 8 time points by iCell8 (wafergen)

I don't know about your experimental design and biological question, but I would just analyse the 8 time points from iCell8 and forget about the Fluidigm. As such you don't have to deal with this batch effect, which is impossible to correct as said before. Afterwards, you can take a look in the Fluidigm data if your results from the iCell8 are confirmed.

[h.mon]

Fluidigm has more than one option of integrated fluidic circuits, and between them, they may have 10x target sequencing depth differences. Maybe you are using one of those circuits?

See this poster: Comparative analysis of single-cell RNA Sequencing Platforms and Methods.

[Za]

Actually Fluidigm C1 is belong to time point 0 when cells are individual cells and start to aggregation at time point 2 hour toward 16 hours (my 8 time points by icell8). Fluidigm C1 data is belong to years ago and boss asked me to combine that with recently sequenced 8 time points in icell8. I have another 9 time points from bulk RNA-seq with the same organism experimental design I also have to compare them with my single cell RNA-seq data (Fluidigm C1 + iCell8 (wafergen)) as ground truth.

[WouterDeCoster]

Then the fluidigm data is completely useless in this story. Not only was it generated on a different platform, it's also from a long time ago, further suggesting that you'll have nasty batch effects.

Analyse everything separately. There is no way you can integrate these three datasets.

[Kevin Blighe]

I concord with Wouter, but best of luck in translating all of what we've said back to your supervisor...