Lab Matters Fall 2024 | Page 99

APHL 2024 POSTER ABSTRACTS investigations identified seven clusters of ≥3 RSV-A infections whose genomes had zero SNP differences , as well as 10 clusters of ≥3 RSV-B infections whose genomes were either indistinguishable or differed only by one SNP . No statistically significant associations between lineages and hospitalization were observed .

Conclusions : We expanded our respiratory pathogen genomic surveillance program to include RSV . Collaboration between laboratory , bioinformatics and epidemiology personnel yielded insights into circulating strains and clusters . We plan to expand sequencing of RSV to include genomes from more adult cases , analyze additional clinical data collected through epidemiologic surveillance and continue to cross-reference sequences against hospitalization data to search for potential genetic markers of virulence .

Presenter : Sean Wang , sean . wang @ state . mn . us

Evaluation of a Bioinformatics Pipeline for Antimicrobial Resistance Prediction in Mycobacterium tuberculosis Complex at New York City Public Health Laboratory

S . Akther , F . Taki , N . Cruz , T . Clabby , J . Wang , A . Khan , G . D . Hawkins , C . Vergara , A . Murthi , M . Chowdhury , A . Olsen , C . Dacosta-Carter , U . Siemetzki-Kapoor , E . Omoregie , S . A Hughes , New York City DOHMH - Public Health Laboratory

The use of whole-genome sequencing ( WGS ) and targeted next generation sequencing ( tNGS ) to detect antimicrobial resistance ( AR ) in Mycobacterium tuberculosis complex ( MTBC ) has substantially increased in recent years . The integration of bioinformatics workflows to detect AR in MTBC in clinical and public health laboratories can substantially improve diagnostic testing and routine surveillance . Here , we report on the evaluation of Bert ’ s bioinformatics pipeline which provides single nucleotide polymorphism ( SNP ) -based AR prediction for MTBC and can be publicly accessed on https :// galaxy . sciensano . be . Bert ’ s pipeline ’ s performance was evaluated using MTBC whole genome sequences ( N &# 3f38 ) that were previously sequenced and analyzed by New York State Public Health Lab , Wadsworth Center ( WC ) and targeted amplicons ( N &# 3f19 ) that were generated by New York City Public Health Laboratory ( NYC PHL ) using tNGS that was adapted from WC . The tNGS approach relies on multiplex PCR and sequencing of 13 targets that have been associated with MTBC resistance to eight antimicrobials . Briefly , the pipeline uses Trimmomatic to trim raw reads , generates quality control ( QC ) reports by FastQC , map reads against MTBC H37Rv reference using Bowtie 2 and estimates coverage using SAMtools . Variants are called , filtered and annotated using BCFtools . Annotated mutations are compared to in-house mutation databases to predict resistance to 17 antimicrobials . We evaluated QC metrics , concordance in AR profiles generated by Bert ’ s and WC ’ s pipelines and analyzed the AR profiles generated for MTBC sequenced at NYC PHL . All 38 WC and 18 / 19 NYC PHL MTBC sequences passed the QC metrics used in Bert ’ s pipeline . Good concordance , 18 / 20 ( 90 %), was observed between Bert ’ s and WC ’ s pipelines for isolates categorized as pan-susceptible by WC . Two genomes categorized as pan-susceptible by WC were identified as mono-resistant by Bert ’ s pipeline . For 18 resistant genomes categorized by WC , 4 / 18 ( 22 %) and 2 / 18 ( 11 %) genomes were found to carry additional high confidence resistant mutations for first-line drugs pyrazinamide ( PZA ) and ethambutol ( EMB ) and second-line drug streptomycin ( STM ), respectively , by Bert ’ s pipeline . Additionally , 9 / 18 ( 50 %) of resistant genomes were found to carry high confidence mutation ( s ) associated with second-line drugs paraaminosalicylic acid ( PAS ), rifabutin ( RBT ) and capreomycin ( CAP ) by Bert ’ s pipeline . These second-line drugs weren ’ t analyzed by WC pipeline . 3 / 18 ( 17 %) of genomes were found to carry additional high confidence resistant mutations for second-line drugs ethionamide ( ETH ) and amikacin ( AMI ) by WC ’ s pipeline , but not by Bert ’ s pipeline . For NYC PHL MTBC sequences , Bert ’ s pipeline categorized 18 and 1 sequences as pan-susceptible and resistant , respectively which were in concordant with phenotypic results . In summary , Bert ’ s pipeline was able to provide QC metrics and produce similar results as WC pipeline . Additionally , Bert ’ s pipeline was able to identify more high-confidence mutations associated with AR and account for more second-line drugs than WC pipeline . This pipeline also generated QC metrics and produced concordant AR profiles for NYC PHL targeted MTBC amplicons . In conclusion , Bert ’ s pipeline generated accurate AR profiles for MTBC using both WGS and tNGS .

Presenter : Saymon Akther , sak-ther @ health . nyc . gov

Evaluating Bioinformatics Approaches for Fungal Pathogen Detection With Targeted Metagenomics

Z . Mudge , J . Mario Vasquez , U . Bagal , L . Gade , S . Morrison , N . Chow , A . Litvintseva , L . Parnell , Centers for Disease Control and Prevention

The spread of fungal diseases is a global public health challenge . Existing laboratory methods used to interrogate the etiology and sources of fungal infections are largely culture-dependent . Advancements in culture-independent techniques , like metagenomic sequencing , provide a promising alternative for more timely and pathogen-agnostic identification of fungal pathogens within environmental and clinical specimens . Specifically , targeted amplicon metagenomic sequencing approaches can provide taxonomic classifications at the genus and species level , but none have been adequately validated for fungal pathogen detection . We previously generated mock fungal DNA communities representing 52 of the most common fungal pathogens to test different primer sets targeting the nuclear ribosomal ITS region . Bioinformatic analysis , using the QIIME2 pipeline and the UNITE database , revealed high classification resolution for specific primer sets . Here , we systematically evaluated additional bioinformatics approaches and databases to identify the optimal analytic methods for resolving the genera and species from the mock DNA communities and to further assess performance of primer sets .

Candidate bioinformatics approaches were divided into two categories : methods and databases . Methods refer to bioinformatics tools or specific uses of tools , while databases refer to the large collections of DNA or ribosomal RNA against which taxonomic classification is run . For this analysis , we ran the nf-core / ampliseq Nextflow pipeline with different parameters that specified combinations of methods and databases . Four taxonomic classification methods were evaluated : DADA2 , QIIME2 , SINTAX and Kraken 2 . Each of these classification methods were run against three different databases : ( 1 ) the UNITE database of the eukaryotic nuclear ribosomal ITS region , ( 2 ) sequences in the NCBI ITS Targeted Loci project and ( 3 ) a custom database specific to fungal pathogens that infect humans . In addition to these tools and databases , we assessed different ITS-specific bioinformatics methods . The variable length of the ITS region can complicate read