Effects of sequence alterations on results from genotypic tropism testing

J Clin Virol. 2015 Apr:65:68-73. doi: 10.1016/j.jcv.2015.02.006. Epub 2015 Feb 10.


Background: geno2pheno[coreceptor] is a bioinformatic method for genotypic tropism determination (GTD) which has been extensively validated.

Objectives: GTD can be affected by sequencing/base-calling variability and unreliable representation of minority populations in Sanger bulk sequencing. This study aims at quantifying the robustness of geno2pheno[coreceptor] with respect to these issues. GTD with a single amplification or in triplicate (henceforth singleton/triplicate) is considered.

Study design: From a dataset containing 67,997HIV-1 V3 nucleotide sequences, two datasets simulating sequencing variability were created. Further two datasets were created to simulate unreliable representation of minority variants. After interpretation of all sequences with geno2pheno[coreceptor], probabilities of change of predicted tropism were calculated.

Results: geno2pheno[coreceptor] tends to report reduced false-positive rates (FPRs) when sequence alterations are present. Triplicate FPRs tend to be lower than singleton FPRs, resulting in a bias towards classifying viruses as X4-capable. Alterations introduced into nucleotide sequences by simulation change singleton predicted tropism with a probability ≤ 2%. Triplicate prediction lowers this probability for predicted X4 tropism, but raises it for predicted R5 tropism ≤ 6%. Simulated limited detection of minority variants in X4 sequences resulted in unchanged predicted tropism with probability above 90% as compared to probability above 98% with triplicate FPRs.

Conclusions: geno2pheno[coreceptor] proved to be robust when sequence alterations are present and when detectable minorities are missed by bulk sequencing. Changes in tropism prediction due to sequence alterations as well as triplicate prediction are much more likely to result in false X4-capable predictions than in false R5 predictions.

Keywords: Geno2pheno; Genotypic tropism determination; HIV-1; Simulation of Sanger bulk-sequencing errors; Triplicate; Tropism.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Base Sequence*
  • Computational Biology / methods*
  • Datasets as Topic
  • Genotype
  • HIV-1 / genetics*
  • HIV-1 / physiology*
  • Phenotype
  • Viral Tropism*