Strenuous efforts were made to compose queries that would capture as many of the references as possible without returning so many spurious hits as to make hand-filtering impossible; in all cases, a single (compound) query was used.

However, conferences (RECOMB and ISMB in particular) are often poorly identified by authors in their citations (many omit the acronym, others mistype it, and those who use the full name of the conference all too often "approximate" it and use a bewildering range of abbreviations for the main words), so that many citations may not be captured by Google Scholar with the query used. (Assuming that Google Scholar does capture all relevant citations, the only thorough search would be one that searches for each article separately, but that would be a full research project in its own right ;-) The problem is compounded in the cases of ISMB and RECOMB by their having had a number of different mechanisms for publishing their proceedings, which has reduced their presence online. Finally, the problem is clearly worst for ISMB, as its proceedings have been published for the last 5 years in the journal Bioinformatics, and many citations to these papers make no reference whatsoever to the conference, treating the citation as a journal article. (Tracking these down would require looking for page numbers that identify supplementary material or querying every article separately.) Thus the values for ISMB, at least, are certainly much too low. (The same problem looms for ECCB, which has followed ISMB's model; in the case of ECCB, it is further complicated by the fact that, for the last 4 years, ISMB and ECCB have been held jointly on alternate years, in which case any references made to conference proceedings typically only bear the ISMB identification.)