Table 1. Break out of 57 surveyed papers in which the authors assess their own methods.

Number of performance metrics	Total number of studies surveyed	Authors' method is the best in all metrics and all data sets	Authors' method is the best in most metrics and most data sets
Note that we did not find any self-assessment paper where the presented method was not top ranked in at least one metric or data set. The survey was conducted over a large pool of scientific peer-reviewed papers selected as follows. First, a Google Scholar search using the keywords ‘computational biology method assessment' was conducted. When papers with comparisons of methods were identified, we further examined (1) papers from the same journal issue and (2) downstream papers that cite the identified paper (as determined by Google Scholar). The 57 papers (see Supplementary information) resulting from the search span 22 journals. Most papers are in the categories of gene regulatory networks/reverse engineering (24/69), structure prediction/assessment (14/69) and DNA–protein interactions/regulatory element identification. An additional nine papers found in the same manner but not shown in the Table reported independently (not-self) assessed methods, of which only four were top performers, whereas five reported methods that ranked high but were not top performers.
1	25	19	6
2	15	13	2
3	7	4	3
4	4	1	3
5	4	1	3
6	2	1	1