Initial interlab analysis is looking great!

[Jake Beal]

Hello, all:

I've performed an initial analysis, which contains all of the data except for Rice replicate #1 (which appears to have only partially uploaded).

For all of the significant fluorescence samples, the geo.std. of group means for the three matched groups is only:

(MG, SG, MR, SR) = 1.3088    2.6344    1.1969    1.1459

when Rice's unmatched flow cytometer is added, these go up to:

1.4765    2.3276    1.4137    1.3791

Meanwhile, the standard deviation between replicates within each group is:

Stanford:      1.1771    1.3559    1.0497    1.0893

Gaitherburg: 1.1854    3.0373    1.1159    1.1247

MIT:              1.0984    2.3766    1.1726    1.1092

Rice:            1.0656    2.0192    1.1000    1.0621

Key takeaways:

1. SG should be discarded as a failed sample: it is all over the place because it contains a mixed population with many non-fluorescing cells.  See attached "SG-debug" comparing 3 Gaithersburg samples.

2. The precision that we are seeing is significantly better than our target: we aimed for 1.5x - 2.0x, and got 1.15x - 1.31x

3. For the most part, inter-group precision is only barely worse than the worst intra-group precision.

Bottom line: as hypothesized, beads appear to be work excellently for comparison, even if we don't exactly know what their units mean!

I look forward to detailed discussion this afternoon, including of the path forward from here.

Thanks,

-Jake