Replicate data can be analyzed individually or aggregated to form a consensus representation of the individual replicate results. The top level analysis option controlling the handling of biological replicates is the "Replicate Data" option. There are two values that can be selected for this option: Individual - Selecting this value will cause replicate data to be processed individually, and each item will appear in the output. Consensus - Selecting this value will cause replicate data to be merged according to the selected resampling method and selected data aggregation mode, and the consensus item will appear in the output.
Identifying Replicate Observation Blocks
Individual observation blocks can be aggregated during the analytical processing to form groups of replicates that are subjected to data merging operations. The groupings are established by the following rules for matching characteristics:
- The individual observation blocks have the same test substance mixtures present in the same degree of dimensionality.
- The individual observation blocks share background treatment (if present) for additional substances not varying on concentration.
- The individual observation blocks share the same protocol name, and the values of the established principal protocol parameters are matching.
Determining Consensus Observation Block Sampling
In the case that the consensus data is requested and replicate data exists for aggregation, the first step in handling each set of replicates is to determine the appropriate concentration sampling for the consensus output. The analysis engine imposes no limitations on the on the merging of replicate data with different concentration sampling. Replicate data can be merged for all cases of heterogeneous sampling amongst replicates. It is important to understand the different resampling algorithms and how they affect the construction of consensus views, including the case where all of the replicates have identical sampling.
The analysis engine provides three resampling algorithms which are selected using the "Resample Method" analysis option. The following table describes the resampling algorithms. The "Resmaple Method" analysis option is only available if the value "Consensus" has been selected for the analysis option "Replicate Data".
|Regular||Produces consensus output with regular log-linear dilution steps.||This algorithm perfoms a fourier signal analysis of the replicates to determine the dominant sampling frequency to be used in constructing the consensus output. The selected frequency is gated by the "Minimum Sample (fold)" analysis option, which is only available if "Regular" has been selected for the "Resample Method" analysis option.|
|Basic||Produces consensus output with irregular, but bracketed dilution steps.||This algorithm arranges all of the distinct concentrations present from all replicates and performs a top down bracketing of concentrations according to the analysis option "Threshold (fold)", which is only available if "Basic" has been selected for the "Resample Method" analysis option.|
|None||Produces consensus output with irregular unbracketed dilution steps.||This algorithm preserves every distinct concentration present in the replicates, no matter how trivial the fold difference between any set of concentrations is.|
Producing Consensus Observation Block Output
Once the resampling has taken place, and the extents of the consensus data item have been established, each location in the consensus output is compared to the available data in the replicates, and any data points that belong in that location are selected and aggregated as either the mean or the median according to the value of the "Data Aggregation Method" analysis option.