Meta-analysis by Information Content (MAIC)
MAIC encodes a simple heuristic to meta-analyse ranked and unranked lists of related named entities. You, the user, decide what is meant by "related": give MAIC a set of input lists, for example experimental results naming genes implicated in a given biological process, and MAIC will iteratively weigh them against each other to learn which lists perform best. The best-performing lists are those that find entities (genes, in our example) that are also found on the other input lists. For a full description of MAIC, see our paper: Li B et al. Nature Communications 11:164(2020)
Enter a set of related lists of named entities that you want to combine to produce a single meta-analysis.
List name: any free-text name to identify this list
Ranked: check this box if the order of items in your list matters. More strongly-supported entities should be listed first.
Category (optional): a single text string (no spaces). If you leave this blank, this input list will be considered to be independent of all other lists. Label lists that come from a similar source by the same category to prevent a given data type from overwhelming the results. For example, "RNA interference".
List contents: paste in the full list of names. Each name must not contain spaces. Separate names by spaces, tabs or commas.
Host factors required for influenza virus replication
In this example analysis, we have aggregated experimental results from a variety of sources relating to host genes involved in influenza virus replication. Data from published sources, together with *unpublished data from Bo Li, JK Baillie and Nir Hacohen.*
Code to run the MAIC algorithm can be downloaded from github.com/baillielab/maic.