Table 3.

Wrongly identified nucleotide sequences summarized according to experimental technique and identity error type.

CorpusTechnique“Non-targeting” yet targeting proportion (%)“Targeting” yet non-targeting proportion (%)Targeting wrong gene/sequence proportion (%)Total per corpus proportion (%)
SGK (n = 115 reagents in n = 75 articles)PCRa0/45 (0)7/14 (50)45/57 (79)52/115 (45)
Gene knockdownb44/44 (100)7/14 (50)12/57 (21)63/115 (55)
Otherc0/45 (0)0/14 (0)0/57 (0)0/115 (0)
Total (Error type)44/44 (100)14/14 (100)57/57 (100)115/115 (100)
miR-145 (n = 49 reagents in n = 31 articles)PCR0/2 (0)8/9 (89)33/38 (87)41/49 (84)
Gene knockdown2/2 (100)1/9 (11)5/38 (13)8/49 (16)
Other0/2 (0)0/9 (0)0/38 (0)0/49 (0)
Total (Error type)2/2 (100)9/9 (100)38/38 (100)49/49 (100)
C + G (n = 109 reagents in n = 51 articles)PCR0/6 (0)23/24 (96)73/79 (93)96/109 (88)
Gene knockdown4/6 (67)1/24 (4)5/79 (6)10/109 (9)
Other2/6 (33)0/24 (0)1/79 (1)3/109 (3)
Total (Error type)6/6 (100)24/24 (100)79/79 (100)109/109 (100)
Gene (n = 284 reagents in n = 128 articles)PCR0/9 (0)35/42 (83)218/233 (94)253/284 (88)
Gene knockdown9/9 (100)7/42 (17)15/233 (6)31/284 (11)
Other0/9 (0)0/42 (0)0/233 (0)0/284 (0)
Total (Error type)9/9 (100)42/42 (100)233/233 (100)284/284 (100)
Oncology Reports (n = 995 reagents in n = 436 articles)PCR0/36 (0)296/335 (88)573/630 (91)869/995 (87)
Gene knockdown30/30 (100)37/335 (11)54/630 (8)121/995 (12)
Other0/36 (0)2/335 (1)3/630 (1)5/995 (1)
Total (Error type)30/30 (100)335/335 (100)630/630 (100)995/995 (100)
Total (n = 1,535 reagents in n = 712 articles)PCR0/89 (0)364/416 (87)937/1,030 (90)1,301/1,535 (84)
Gene knockdown87/89 (98)50/416 (12)89/1,030 (9)226/1,535 (15)
Other2/89 (2)2/416 (1)4/1,030 (1)8/1,535 (1)
Total (Error type)89/89 (100)416/416 (100)1,030/1,030 (100)1,535/1,535 (100)
  • a PCR = Human gene or genomic targeting primers for PCR, RT–PCR or methylation-specific PCR.

  • b Gene knockdown = siRNA or shRNA.

  • c Other = Claimed Ribozyme, TALEN, mimic sequences, and other oligonucleotide sequences.

  • Bold text indicates the most frequent error types per corpus.