Catching up on 🧬 “Understanding #NGS and Interpreting Reports for Oncologic Pathology” with Dr. @sabahkadri and Dr. Cecilia Yeung at #CAP21 @AMPath 👩🏻💻
Super useful talk with excellent case studies! Here are just a few takeaways 🧵⬇️ #MolecularPath_CEJ #Informatics_CEJ
Super useful talk with excellent case studies! Here are just a few takeaways 🧵⬇️ #MolecularPath_CEJ #Informatics_CEJ
🤔 Considerations in developing an NGS assay
📌 @Pathologists has NGS assay development tools regarding test design, validation, quality, and bioinformatics: cap.org/member-resourc…
📌 How much do you want to know? ➡️ what test is needed? (WGS vs WES vs targeted sequencing)
1/
📌 @Pathologists has NGS assay development tools regarding test design, validation, quality, and bioinformatics: cap.org/member-resourc…
📌 How much do you want to know? ➡️ what test is needed? (WGS vs WES vs targeted sequencing)
1/
🧬 NGS in a nutshell with Dr. Yeung
1️⃣ Library preparation (hybrid capture vs amplicon-based)
2️⃣ Sequence the fragments of DNA (cyclic reversible termination vs detection of chemical byproducts ⚡️ as single nucleotide is added)
2/
1️⃣ Library preparation (hybrid capture vs amplicon-based)
2️⃣ Sequence the fragments of DNA (cyclic reversible termination vs detection of chemical byproducts ⚡️ as single nucleotide is added)
2/
Other NGS techniques:
📌 Multiplexing samples: tag each library component from the same patient with indentical barcode ➡️ combine patient samples into one sequencing reaction ➡️ de-multiplex with #bioinformatics after sequencing
3/
📌 Multiplexing samples: tag each library component from the same patient with indentical barcode ➡️ combine patient samples into one sequencing reaction ➡️ de-multiplex with #bioinformatics after sequencing
3/
🏷 Unique molecular identifiers (UMI)
📎Unique to each library component
📎 Allows you to distinguish noise from true variants (if all reads with same UMI have a change ➡️ likely true variant; if only some ➡️ likely noise)
4/
📎Unique to each library component
📎 Allows you to distinguish noise from true variants (if all reads with same UMI have a change ➡️ likely true variant; if only some ➡️ likely noise)
4/
🔎 Analysis
1️⃣ Alignment to reference sequence
2️⃣ Variant calling and quality analysis
3️⃣ Annotation ✍️
4️⃣ Report
5/
1️⃣ Alignment to reference sequence
2️⃣ Variant calling and quality analysis
3️⃣ Annotation ✍️
4️⃣ Report
5/
📝 Navigating NGS reports
Variants are classified into 5 categories:
📌 Benign 👍🏼
📌 Likely benign 🤔
📌 Uncertain 🤷🏻♀️
📌 Likely pathogenic 😣
📌 Pathogenic 😱
6/
Variants are classified into 5 categories:
📌 Benign 👍🏼
📌 Likely benign 🤔
📌 Uncertain 🤷🏻♀️
📌 Likely pathogenic 😣
📌 Pathogenic 😱
6/
Or looked at another way...
📌 Tier I: strong clinical signficance (therapeutic, diagnostic, prognostic w level A or B evidence)
📌 Tier II: potential clinical signficance (⬆️ but w level C or D evidence)
📌 Tier III: unknown clinical signficance
📌 Tier IV: B9 or likely B9
7/
📌 Tier I: strong clinical signficance (therapeutic, diagnostic, prognostic w level A or B evidence)
📌 Tier II: potential clinical signficance (⬆️ but w level C or D evidence)
📌 Tier III: unknown clinical signficance
📌 Tier IV: B9 or likely B9
7/
👩🏻💻 Bioinformatics (yay! 😍) with @sabahkadri :
📌 How we convert data from sequencers into useful information for pathologists to interpret and pass along to clinical colleagues through our reports
8/
📌 How we convert data from sequencers into useful information for pathologists to interpret and pass along to clinical colleagues through our reports
8/
🌾 Problem scope: looking for a genomic anomaly (needle) in the whole genome (haystack) ➡️ but bioinformatics is your tool! (maybe a metal detector??)
9/
9/
#Bioinformatics pipeline:
📌 Black box software vs developing a custom pipeline with quality assessment, alignment, variant calling
📌 Transformation of data in pipeline: raw data (as FASTQ) ➡️ alignment (as SAM/BAM) ➡️ SNVs and indels (as VCF)
10/
📌 Black box software vs developing a custom pipeline with quality assessment, alignment, variant calling
📌 Transformation of data in pipeline: raw data (as FASTQ) ➡️ alignment (as SAM/BAM) ➡️ SNVs and indels (as VCF)
10/
🔎 Visualize the data with an integrated genomics viewer:
software.broadinstitute.org/software/igv/M…
11/
software.broadinstitute.org/software/igv/M…
11/
📍 At this point, ready for tertiary analysis (variant annotation) with software (e.g. ANNOVAR) annovar.openbioinformatics.org/en/latest/
1️⃣ Determine the type of variant, 2️⃣ compare using databases (COSMIC, ClinVar), and then 3️⃣ filtering and ranking variants
12/
1️⃣ Determine the type of variant, 2️⃣ compare using databases (COSMIC, ClinVar), and then 3️⃣ filtering and ranking variants
12/
Thank you for this excellent talk Dr. @sabahkadri and Dr. Yeung!
⬆️ is just a brief summary of a few points - highly recommend viewing the full talk on @Pathologists @AMPath #CAP21, which also has great case studies to illustrate these concepts in action
End 🧵
⬆️ is just a brief summary of a few points - highly recommend viewing the full talk on @Pathologists @AMPath #CAP21, which also has great case studies to illustrate these concepts in action
End 🧵
@threadreaderapp unroll
• • •
Missing some Tweet in this thread? You can try to
force a refresh
