IHMT

Instituto de Higiene e Medicina Tropical

Home / Publicações / Evaluation of the automatic editing tool RECall for HIV-1 pol and V3 loop sequences

Evaluation of the automatic editing tool RECall for HIV-1 pol and V3 loop sequences

Genotypic drug resistance testing is routine practice in HIV-1 clinical care. The visual interpretation of sequencing electropherograms is labour-intensive and subject to intra- and inter-assay variability because decisions are based on operators’ judgments. In this study the performance of the automatic editing tool RECall was compared to the current standard of editing manually and editing using the tool ViroSeq.

Using RECall a consensus sequence could be generated for 97% of the V3 loop and for 79% of the pol experiments. By comparison, using manual editing a consensus sequence could be reached for 87% of the V3 and 87% of the pol experiments. Using ViroSeq, a consensus sequence was generated for 68% of the pol experiments. On a predefined dataset, manual editing displayed the highest probability to accurately assign mixtures (0.91 vs. 0.88 by ViroSeq vs. 0.76 by RECall) and the lowest probability to inaccurately assign a mixture to a pure base call (0.002 vs. 0.019 by ViroSeq vs. 0.002 by RECall). As differences in base calling have little impact on drug resistance interpretation and hands-on-time could be substantially reduced, RECall could be a valuable tool for the standardization and acceleration of the editing process.

Evaluation of the automatic editing tool RECall for HIV-1 pol and V3 loop sequences

Genotypic drug resistance testing is routine practice in HIV-1 clinical care. The visual interpretation of sequencing electropherograms is labour-intensive and subject to intra- and inter-assay variability because decisions are based on operators’ judgments. In this study the performance of the automatic editing tool RECall was compared to the current standard of editing manually and editing using the tool ViroSeq.

Using RECall a consensus sequence could be generated for 97% of the V3 loop and for 79% of the pol experiments. By comparison, using manual editing a consensus sequence could be reached for 87% of the V3 and 87% of the pol experiments. Using ViroSeq, a consensus sequence was generated for 68% of the pol experiments. On a predefined dataset, manual editing displayed the highest probability to accurately assign mixtures (0.91 vs. 0.88 by ViroSeq vs. 0.76 by RECall) and the lowest probability to inaccurately assign a mixture to a pure base call (0.002 vs. 0.019 by ViroSeq vs. 0.002 by RECall). As differences in base calling have little impact on drug resistance interpretation and hands-on-time could be substantially reduced, RECall could be a valuable tool for the standardization and acceleration of the editing process.