This section represents what is perhaps the most important part of the TMA-Combiner. At first glance, one may think that combining scores by simple averaging would be sufficient. However, when it comes to combining stained and scored IHC replicate cores, the situation is not that simple! One needs to take a step back and carefully consider the significance of what the combination process is doing to these IHC-stained replicate cores. Issues such as core sampling biases, antibody staining properties, and number of replicate cores weigh considerably in this process.

We discuss the importance of these considerations and why we constructed the score combination rules shown below, in our publication, so we won't repeat it here. We will, however, go into greater detail on the mechanics of the score combination process, particularly on how each rule determines the outcome score from the input score pool of replicate cores.

First off, it is important to establish the scoring system that is introduced earlier in the TMA-Deconvoluter walkthrough. The TMA-Combiner can handle quantitative scoring systems, which we describe in greater detail here. However, we use a discrete integer scoring system, so the TMA-Combiner defaults to that, as shown in the score key below:

† Background colors indicate how they would appear in TreeView

Note that scores are either user scores or TreeView scores. This is dependent on the type of output selected for the TMA-Deconvoluter - PCL files use TreeView-compatible scores, whereas K-M files use unconverted, user scores. User scores might also be present if the user generated the input files from other sources or with methods other than the TMA-Deconvoluter.

In order for the TMA-Combiner to apply the score combination rules correctly, the user will need to indicate whether or not user scores are present in the input files. Since the PCL file format is the native format for the TMA-Combiner, the TMA-Combiner will assume by default that the input files contain TreeView-compatible scores. Thus, if the user is combining scores based on the user's scoring system, the user should select the option below:

**In the case of missing datapoints, while the default is a blank cell, the user may use a number or symbol (such as "X") instead. This should be indicated in the Score Key, like in the example below:

One other, very important consideration with scores – if the user is planning to combine multiple TMAs together into a single file, the user should ensure that every file in the entire dataset consists entirely of user scores or of TreeView-compatible scores (i.e. the constituent files comprising the whole dataset to be combined should contain only one type of score, not both). Otherwise, score combination might not be performed correctly and may produce errors. The easiest way to fix this is to use the TMA-Deconvoluter's Score Conversion Utility to convert the scores. For more information on this, consult the TMA-Deconvoluter documentation.

Back to Step 1
Step 3 - Score Combination Rules, Part 2