Textrous! is a free web framework designed to automatically retrieve English words from a gene set. Textrous! aims to search "gene-documents", collections of genes and their associated literature based upon databases from PubMed, Jackson Laboratories, and the Online Mendelian Inheritance in Man. The algorithm behind Textrous! is latent semantic indexing, a technique used to discover the semantic structure of documents by examining statistical co-occurence patterns. As a result, Textrous! is capable of extracting both direct and indirect links between genes and words.
- Different Methodologies - We are capable of processing gene sets in two different ways: collectively and individually.
- Collective Processing - The features of every gene are combined into an "average" gene. Advantages: Nothing is discounted. Words that are not statistically significant for each respective gene may be significant for the gene set. Disadvantages: Sensitivity to outliers.
- Individual Processing - Every gene-word association is computed individually. Advantages: Outliers are accounted for. Disadvantages: Gene independence is assumed, which may not be true.
- Data Tables - Shows the top related words to the collectively processed gene set, and their associated cosine similarity, z-score, or p-value.
- Hierarchical Cloud - Shows the top related words to the collectively processed gene set in a word cloud/tree hybrid. Words are clustered with an agglomerative hierarchical clustering algorithm.
- Heat Map - Shows the top related words to each individually processed gene. White cells and colored cells represent statistical non-significance and significance, respectively.
- Phrasing - Shows the top phrases associated with a word, sorted by their association with your gene set. Only noun phrases are supported.
Textrous! uses a list of words excluded from the searching algorithm. To view the list, click here.
Current versions of the following: Firefox, Google Chrome, Internet Explorer, Safari, Opera.
Copy-and-paste from Excel spreadsheets may not function with Internet Explorer.