P22-19. MiDRMpol: A simplified, cloud based algorithm to identify minor HIV-1 drug resistance mutations

Background
The detection and quantification of minor populations of drug resistant mutations (DRM) in viral quasispecies is a major research interest in the HIV-1 resistance field. However high throughput sequencing (HTS) and implementaion of computationally intensive bioinformatics pipeline still remain a challenge. The aim of the present study is to develop a computationally efficient algorithm, which takes fastq files generated by HTS-platforms and deliver easily inferable diagnostic report and high quality visualization of the results. The algorithm executes in cloud-based environment in order to become accessible globally through internet without advanced computational resources.

Methods
Total of 87 HIV-1 positive plasma samples from both treatment naïve and experienced patients with viral load >2000 copies/ml were used to amplify the gag-pol. The amplified regions were sequenced in Illumina HiSeq2500 platform after pooling 48 samples together in one run. The resultant raw paired-end reads were pre-processed and analysed using automated bioinformatics pipeline. Open-source Linux based tools and in-house Perl and Python programs were used for the analysis. The identified mutations were compared and infered with Stanford HIV Drug Resistance Database followed by visualization in CIRCOS plot. The DRM by GRT-PS was performed using ViroSeq.

Results
The high throughput sequencing identified 14% (7/50) of treatment naïve individuals as having sDRM in Swedish Cohort, of whom two (4%) had only major-PI mutations (M46I), three (6%) had only NRTI mutations (M184I, T215S and K219R) and two (4%) has major-INI mutations (Q148H and E138K) (Figure 1). In our Ethiopian cohort, six out of 17 patients has DRM of whom five had single class mutations and one had both single NRTI (M184I) and INI (Q148H) mutations. In our Indian cohort, five of 10 samples had sDRM. Interestingly all five of the latter had acquired infection vertically through mother-to-child transmission. The assay can detect minor viral population at 1% level. In the entire three cohorts, MiDRMpol identified additional major resistance mutations against all three drug classes at minor viral quasispecies while comparing with ViroSeq.

Conclusions
MiDRMpol workflow is a cost as well as computational effective method for employing high-throughput sequencing to identify the minor population of drug resistance (<20%) mutations in protease, reverse transcriptase and integrase inhibitors. This method can be implemented without prior knowledge in the HTS. The end user can upload the FASTQ file as input and obtained the report with DRMs at different frequency.