blairlearn
commented
3 years ago Test cases:
- steroid - matches on word boundary, search will find matches not on word boundary.
- anti-c - matches on word boundary, variations will match or fail to match "anti c" (with a space).
| Top 50 drug search terms Search Term | Searches |
|---|---|
| nivolumab | 718 |
| pembrolizumab | 669 |
| bevacizumab | 532 |
| atezolizumab | 511 |
| durvalumab | 444 |
| all | 427 |
| trastuzumab | 411 |
| avelumab | 383 |
| ipilimumab | 364 |
| olaparib | 353 |
| cetuximab | 340 |
| trametinib | 334 |
| palbociclib | 313 |
| paclitaxel | 293 |
| cabozantinib | 291 |
| cyclophosphamide | 274 |
| alpelisib | 272 |
| carboplatin | 267 |
| cisplatin | 263 |
| abemaciclib | 261 |
| selumetinib | 259 |
| venetoclax | 256 |
| regorafenib | 256 |
| ramucirumab | 245 |
| sorafenib | 243 |
| tremelimumab | 241 |
| rituximab | 238 |
| everolimus | 236 |
| cobimetinib | 235 |
| lenvatinib | 234 |
| gemcitabine | 231 |
| capecitabine | 229 |
| niraparib | 227 |
| doxorubicin | 226 |
| sunitinib | 225 |
| lenalidomide | 221 |
| bortezomib | 214 |
| folfox | 213 |
| binimetinib | 211 |
| crizotinib | 211 |
| daratumumab | 203 |
| docetaxel | 203 |
| lapatinib | 202 |
| irinotecan | 198 |
| selinexor | 196 |
| afatinib | 195 |
| pdr001 | 194 |
| erlotinib | 191 |
| ibrutinib | 187 |
| entrectinib | 185 |
| erdafitinib | 185 |
Issue description
Autocomplete queries for drug terms containing a string return "unusual" matches. For example, searching for drug terms containing the "anti-c" should only terms with strings matching on "anti-c" when it starts on a word boundary. An expected result would be Term 802323: "allogeneic anti-CD20 CAR T cells LUCAR-20S"
Instead, there are also matches with inexact matches: e.g. Term 764075: :anti c-KIT antibody-drug conjugate LOP628" (matches on "anti c" with a space instead of a dash) Term 794677: "anti-EGFR monoclonal antibody CPGJ 602" which has the text "EGFR monoclonal antibody " between the "anti" and the "c".
Term 764075 might be expected because of ignoring punctuation, but 794677 doesn't fit expectations.
Similar unexpected results occur when searching for drug alias records.
Steps to reproduce the issue
https://webapis.cancer.gov/drugdictionary/v1/Autosuggest?searchText=anti-c&matchType=contains&size=20&includeResourceTypes=DrugTerm
What's the expected result?
What's the actual result?
Additional details / screenshot
Related Tickets