EBBS: Error Based Blast Sequence filter
Description:
We aim to provide guidance on what confidence thresholds to use for taxonomic assignment of unknown sequences from marine environments using BLAST. This R-Shiny application is associated with the publication by Pappalardo et al. in Methods in Ecology and Evolution. Metabarcoding/eDNA techniques are a fast and cost-effective method that allows identifying organisms in a mixed community or environmental sample based on information from a genetic marker. BLAST is the most popular alignment-based tool that measures how similar the unknown sequence is to the sequences in the DNA reference database. In addition to filter blast results for some minimal quality, researchers often use a confidence threshold for all the dataset to decide which matches to trust to phylum or species level (we refer to this as *global threshold*). Occasionally, researchers may use intermediate thresholds for different taxonomic levels. However, due to different evolutionary rates across different taxa, we would also expect that the optimal threshold for taxonomic assignment would vary depending on the taxonomic group. Pappalardo et. al. developed taxon-specific *optimized thresholds* for taxonomic assignment - with a focus on marine taxa that are currently underepresented in reference databases, also testing two methods for selecting the BLAST final match. This app allows users/researchers to input their BLAST results, define their parameters for quality filtering, pick a method to select the final hit (best-hit or best-shared), define which thresholds to use (global or optimized) and their parameters, and input their desired error rates (if using the optimized thresholds option). We provide suggested defaults for all options. Finally, users can download their final filtered results. OI is one of the most widely used genetic markers for targeting metazoans in barcoding and metabarcoding applications. The current version of the app provides optimized thresholds only for COI data, but we plan to expand in a future release.
Acknowledgments:
Pappalardo et al. work and code development were funded through a Smithsonian fellowship by grant NMNH ASCS Core Research Grant 2021 awarded to Karen Osborn. Development of the R shiny app was done by Matthew Kustra (matthewckustra@gmail.com) based on R code written by Paula Pappalardo.
Disclaimer:
We did our best to provide a web application that can implement the approach recommend by Pappalardo et al. (Methods in Ecology and Evolution). Please note that this application is provided as is with no guarantees of completeness and accuracy. The authors of the publication and the app developer assume no responsibility and shall not be liable for unintended results.
