Skip to content

Spkinein normalization

If you are using euromrd spike-in protocol, you can use our spike-in configurations, that enriches the .vidjil file with information concerning MRD.

You can select one of the two configurations, depending on your preference:

  • median method: the normalization factor for each locus is computed using the median of spike normalization factors. Outliers are kept in the computation, but median is a computation method which is robust to outliers. If there were outliers in your sample, they are still detected and logged for information purposes.

  • mean method: the normalization factor for each locus is computed using the mean of spike normalization factors, excluding outliers from this computation.

For each method, a locus normalization factor is computed if there were at least one spike for this locus in the sample. For each spike, the normalization factor corresponds to an amplification factor: it is computed as the number of reads of the spike found in the sample divided by the number of copies of the spike you put in the sample.

These values are used to compute several metrics that will be displayed in the client.

Spikes⚓︎

  • visualize your spikes directly in the graph: they are tagged with the standard/spike label, which appears in light blue in the different parts of the Vidjil window (sample graph, plot view and clone list).

  • find some specific warnings about the spikes :

    • if a spike was expected but not found in the sample (it is generated with 0 reads and the warning "W32 - Spike not detected in sample (generated with 0 reads)" should appear in the warnings list)
    • if a spike is over-represented or under-represented regarding the other spikes in the sample, it is classified as an outlier and is excluded from locus normalization factor if you chose the mean method configuration. A warning should appear both in spike clone info and in the warning list ("W31 - Outlier spike: normalization factor anomaly detected")

  • In the plot view, you can find a dedicated preset "spike distribution" that allows you to see the distribution of spike normalization factors.

Clones informations⚓︎

In the clone information window (little "i" on the right end of the clone line, "clonotype information"), a new subsection in the table called "MRD normalization" is displayed. The information you can find in this table depends on the type of the selected clone:

  • if the clone is a spike, you can find this particular spike's normalization factor and the locus normalization factor, both for each time point.
  • if the clone is not a spike, you can find the locus normalization factor used to compute the normalized values for this clone, and the two normalized values:
    • normalized cells, corresponding to the estimated number of cells of this clone in the sample
    • normalized reads, corresponding to the estimated number of reads of this clone regarding the amplification factor.

In the clone list, you can also change the values displayed in the right column. By using the "ctrl+A" shortcut, you can change the type of values you want to see, including spike factor, normalized cells and normalized reads values.

Sample informations⚓︎

In the sample information window (little "i" on the left of the sample line, "sample information"), a dedicated "MRD" table is displayed.

In this table, you can find:

  • sensitivity and sample cell count information: the sample cell count corresponds to the number of cells effectively found in the sample, minus the spike cells, and normalized with a global normalization factor (with all spikes counted together); the sensitivity is then 1 divided by the sample cell count.

  • a subtable containing information about the normalization factors: for each locus with a computed normalization factor, you can find its value, the lower and upper bounds computed to detect outliers and the list of individual spike normalization factors for the locus.

Settings menu - normalization⚓︎

In the settings menu, you have now a "MRD" normalization button (only if you used a MRD configuration and you have MRD data in your output). This option recomputes the size of the clones with the "normalized reads" value. In this case, the spikes are excluded from the computation, so the percentage takes into account only the natural clones.