-
Notifications
You must be signed in to change notification settings - Fork 18
3. Examples and Exercises
This section provides scenarios that demonstrate various capabilities of the FHIR Genomics Operations. You can also explore these scenarios using the postman collection.
We are interested in patient HG00403's LDLR variants. Let's look at genomic liftover, and see if we get the same results by querying with a build37 (NC_000019.9:11200138-11244496) vs build38 (NC_000019.10:11089431-11133820) range.
- returns 20 variants
- returns 20 variants
What if we're only interested in, say, LDLR Exon 7 (of MANE transcript NM_000527.5) (NC_000019.9:11221327-11221447)
- returns 0 variants
Exercise
- We want to know the distribution of the 20 variants across LDLR exons. Develop an application that calls $find-subject-variants for each LDLR exon, and displays the count of results in a histogram.
We are interested to know if patient m123 has a particular APC variant. The lab that reported the variant only reports transcript level variants, not necessarily based on the MANE transcript. Let's look at normalization, and see if this particular variant is present, searching with any HGVS or SPDI synonym from the ClinVar page for this variant.
- returns variant dv-639224f104908c0632d26136
- returns variant dv-639224f104908c0632d26136
- returns variant dv-639224f104908c0632d26136
Exercise
- Would you expect the find-subject-variants operation to also encapsulate liftover, in addition to normalization? For instance, what variant would you expect to be returned if you search for NC_000005.10:g.112766335A>T?
- Patient NA19238 is the mother, and patient NA19239 is the father of patient NA19240. Patient NA19240 is found to have rs112747179 (NC_000009.11:g.35854651GT[10]). Was this variant inherited from mother or father or is the variant de novo in the patient?
It should be noted that normalization and liftover sometimes fail (e.g. NCBI SPDI services may not normalize a particular variant, for various reasons). For instance, we now want to know if patient m123 has a particular SAMD11 variant, but we don't know how it was reported by the lab. HGVS and SPDI synonyms are on this ClinVar page. We can find the variant using a canonical SPDI or 'NC_' HGVS, but we get an error when searching with 'NM_' HGVS expressions:
- returns variant dv-639224f104908c0632d26135
- returns variant dv-639224f104908c0632d26135
- ERROR 400: Cannot normalize NM_001385641.1:c.804C>T
- (Side note - this now DOES return dv-639224f104908c0632d26135, showing that NCBI SPDI services are constantly improving!)
Exercise
- Variants in genes other than LDLR are also associated with familial hypercholesterolemia. Does patient HG00403 have this particular APOB variant?
- In the case of a failed liftover, how might a client alter the query in order to be successful?
We now want to compute a polygenic risk score on patient HG00404. Our simple (fake) score is based on the presence or absence of 7 SNPs, where we count '1' for each SNP present, and the total score is the sum. The higher the sum, the greater the risk for a hereditary condition.
- None are present. Score=0
Patient HCC1143 has a tumor recurrence. WES of new mass (specimen ID = 'HCC1143-Sp1') is performed. The patient and their clinician are considering therapeutic options, including a new clinical trial with inclusion criteria of any of the following: MET amplification, EGFR wildtype, BRAF class 2 mutation, KIT exon 11 mutation. Does patient HCC1143 have any structural variants in this specimen that overlap EGFR (NC_000007.14:55019016-55211628) or KIT exon 11 (NC_000004.12:54727415-54727542)?
- Based on the results of this query, what can you conclude about patient satisfying EGFR or KIT inclusion criteria?
Exercise
- BRAF class 2 mutations include [‘NC_000007.14:140781616:C:T’, ‘NC_000007.14:140781616:C:A’, ‘NC_000007.14:140781601:C:G’, ‘NC_000007.14:140781602:C:T’, ‘NC_000007.14:140781601:C:A’, ‘NC_000007.14:140753333:T:C’, ‘NC_000007.14:140753331:T:G’, ‘NC_000007.14:140753332:T:G’ ,’NC_000007.14:140753344:A:T’, ‘NC_000007.14:140753345:G:C’]. What operation would you use to see if any are present in patient HCC1143?
Patient ABC789 has an idiopathic dilated cardiomyopathy. The most frequent genetic cause is a variant in the A-band (NC_000002.12:179400709-179483218) of the TTN gene. Prior investigations have not identified any simple variants in this region, and a researcher now wants to assess for possible overlapping structural variants.
- A deletion is found
Patient HCC1143 has a tumor recurrence. WES of new mass (specimen ID = 'HCC1143-Sp1') is performed. The patient and their clinician are considering therapeutic options, including a new clinical trial with inclusion criteria of any of the following: MET amplification, EGFR wildtype, BRAF class 2 mutation, KIT exon 11 mutation. Does patient HCC1143 have any structural variants in this specimen that fully subsume MET (NC_000007.14:116672195-116798386)?
- Based on the results of this query, what can you conclude about patient satisfying MET inclusion criteria?
Patient NB6TK328 underwent preemptive pharmacogenomic testing. Clinician now wants to prescribe amitriptyline, but first wants to see CYP2D6 and CYP2C19 star alleles.
- Results show CYP2D6 *1/*41 and CYP2C19 *1/*17
Exercise
- What operation could you use, to determine the implications of these star alleles on the use of amitriptyline?
Operations having coded elements as input parameters can accept a token or a codesystem|code as described here. Let's assume that patient NB6TK328 has previously undergone WES for another indication, and is now experiencing a gradually progressive dull pain in the lower back, associated with morning stiffness. Patient's clinician is suspicious for ankylosing spondylitis, and wants to see the patient's HLA-B alleles to rule out HLA-B27. Try setting the 'genes' parameter to 'http://www.genenames.org/geneId|HGNC:4932', 'HLA-B', 'HGNC:4932', '4932' to convince yourself that you get back the same results each time:
- Patient's genotype is HLA-B08:01/HLA-B40:01
- Patient's genotype is HLA-B08:01/HLA-B40:01
This is the same scenario as above, but now, for patient NB6TK328, we will search for specific HLA-B haplotypes.
- Nothing is found (as expected)
Exercise
- Knowing that NB6TK328's HLA-B genotype is HLA-B08:01/HLA-B40:01, show how you can use the find-subject-specific-haplotypes, searching by either a token (e.g. 'HLA-B*08') or a codesystem|code, to confirm.
Patient NB6TK328 underwent preemptive pharmacogenomic testing. Clinician now wants to prescribe amitriptyline, but first wants to check for any drug-gene interactions.
- Results show CYP2D6 *1/*41 (normal metabolizer) and CYP2C19 *1/*17 (rapid metabolizer)
We learned from the find-subject-haplotypes operation that patient NB6TK328 has CYP2D6 *1/*41 and CYP2C19 *1/*17 genotypes. Patient's clinician now wants to know ALL medications whose metabolism is potentially affected by these genotypes.
- the metabolism of many medications are affected
Patient CA12345 has metastatic non-small cell lung cancer. Biopsy shows two somatic variants felt to be oncogenic: NM_002524.5:c.182A>C (NRAS:p.Gln61Pro), NM_001354609.2:c.1799T>A (BRAF:p.V600E). Clinician now wants to determine if there are any molecularly-guided medication treatment options for this patient.
- CIViC data shows resistance to dabrafenib, sensitivity to vemurafenib, and sensitivity to trametinib+dabrafenib
A family member of patient huC30902 carries a presumptive diagnosis of Hereditary Paraganglioma-Pheochromocytoma Syndrome (https://www.ncbi.nlm.nih.gov/medgen|C1708353). Does patient huC30902 have any variants associated with this condition?
- Two variants are found
Exercise
- What are the clinical significances of the identified variants?
- The reference implementation returns the ClinVar Variation ID of variants found in the patient that are also present in ClinVar. Find the ClinVar Variation IDs in the operation response, and then look up the variants in ClinVar. What is the review status shown in ClinVar? Is this review status also present in the operation response?
The CDC defines Tier 1 variants as those having significant potential for positive impact on public health based on available evidence-based guidelines and recommendations. Tier 1 variants include those causing Hereditary Breast and Ovarian Cancer Syndrome (HBOC), Lynch syndrome (LS), and Familial hypercholesterolemia (FH). Does patient NB6TK329 have any variants associated with these conditions?
- Does patient have any variants associated with these conditions, other than those classified as benign?
Exercise
- What are the genes associated with CDC Tier 1 variants? Is there an operation you can use to specifically look for the presence of variants in any of these genes?
Patient HG02657 has liver disease, and the patient's clinician suspects hemochromatosis. The clinician submits a search for variants associated with hereditary hemochromatosis
- One variant is found
Exercise
- Examine the corresponding ClinVar record. Given the conflicting interpretations of clinical significance, how would you make a judgment in this patient's case?
A patient with lung cancer has no known druggable targets. Their clinician wants to see if any variants in ALK (NC_000002.12:29190991-29921589), BRAF (NC_000007.14:140713327-140924929), or EGFR (NC_000007.14:55019016-55211628) are predicted to be consequential.
- Multiple consequences are found
The clinician wants to further constrain the query in the prior example to only return variants predicted to affect the 3' untranslated region.
- Response is filtered to just those variants that affect the 3' UTR.
Patient HG00403 is detected as having a variant (NC_000001.10:86852620:A:G) in the ODF2L gene, and their clinician wants to further characterize the variant by examining its predicted molecular consequences.
- What is the most significant consequence of the stated variant?
Operation input parameters can be arranged to represent 'AND' vs 'OR' logic as described here. In this case, a clinician discovers that their patient NA19238 has a cluster of variants in the q21.1 region of Chr1, and wants to know if there are other patients that share ALL of these variants:
- Six patients share ALL of these variants
To further investigate, the clinician wants to know if there are other patients that share ANY of these variants:
- Additional patients are identified
Exercise
- Let's revisit the topic of normalization. Operations encapsulate normalization, freeing you from having to search for all the different ways a variant might be encoded. There are several patients in our population that have pathogenic LDLR variant NC_000019.10:11089559:G:A. Searching against non-normalized variant observations in a standard FHIR server (https://api.logicahealth.org/FHIRgenomics2Test/open/), determine how many patients have this variant. Having done that, determine how many patients have this variant using the find-population-specific-variants operation. Do you get the same count? Which method is easier? Which method is more accurate?
A researcher has developed a new drug, designed for cancer patients with large deletions involving all or part of BRCA1 (NC_000017.11:43044294-43125364) or BRCA2 (NC_000013.11:32315507-32400268), and wants a list of potential clinical trial participants.
- 0 patients with BRCA1, 1 patient with BRCA2 intersecting structural variants
Exercise
- What operation can you use, to more closely examine the patient's BRCA2 structural variant?
A researcher has developed new drugs, designed for cancer patients with copy number increases of ERBB2 (NC_000017.10:37844346-37884911) or MET (NC_000007.13:116312249-116438431), and wants a list of potential clinical trial participants.
- 0 patients with ERBB2, 1 patient with MET subsuming structural variants
Exercise
- What operation can you use, to more closely examine the patient's MET structural variant?
A pharmacogeneticist is studying the accuracy of CYP2D6*5 (whole gene deletion) calling, and wants to compare cases where a structural variant caller indicates a whole gene deletion of CYP2D6 (NC_000022.10:42522500-42526812) against a pipeline used to report pharmacogenes. To begin the process, the pharmacogeneticist identifies all patients that have structural variants subsuming CYP2D6.
- 1 patient
Exercise
- What operation can you use, to more closely examine the patient's CYP2D6 structural variant?
A pharmacist is interested in CYP2C19 metabolic effects, and wishes to identify all patients having been genotyped for this gene.
- 12 patients
A transplant specialist is looking for potential unrelated donors in the donor registry that are a perfect HLA match for their patient. Are there any patients that have ALL of these haplotypes: HLA-A01,HLA-B08,HLA-DRB1*03?
- 1 match is found
Exercise
- How would you alter the query, to find patients that have ANY of these haplotypes: HLA-A01,HLA-B08,HLA-DRB1*03
A pharmacist wants to examine the prevalence of altered amitriptyline metabolism in a population, and issues a query for all patients having any type of tx implication to amitriptyline.
- 15 patients are found
A health plan establishes a quality assurance program aimed at ensuring that patients with NSCLC who are candidates for molecularly guided treatment are properly identified. To investigate, the health plan queries a repository to obtain a list of patients that have a therapeutic implication involving NSCLC, irrespective of variants/haplotypes/genotypes from which the implication was derived.
- 15 patients are identified
Exercise
- Note that this search retrieves both somatic and germline variants. Take a closer look at patient NA18498 - what variant is associated with a tx implication? Use the variant's CIViC variant ID to look up the specific variant in CIViC.
Over the course of the past two years, many patients have been tested for variants pathogenic for hereditary breast and ovarian cancer syndrome (HBOC), Lynch syndrome (LS), and familial hypercholesterolemia (FH). Recent knowledge base updates have added to the list of known pathogenic variants, and have reclassified the pathogenicity of many variants. We now want to find patients that have pathogenic variants for HBOC, LS, or FH, based on the new knowledge base.
- 19 patients
As part of a research protocol designed to build better oncogenicity prediction algorithms, an oncologist seeks to identify patients harboring intron variants.
- 16 patients
Feb 2020, patient HCC1143 presents with headache, and mass is found on head CT. Biopsy is consistent with metastatic lung cancer. WGS is performed on the biopsy specimen, and analyzed for simple and structural variants. For technical reasons, portions of the genome were deemed uncallable. To be eligible for a particular clinical trial, patients must have an EGFR (NC_000007.14:55019016-55211628) exon 19 (NC_000007.14:55174721-55174820) deletion. Patient's oncologist finds no simple or structural variants in the region of EGFR exon 19, and asks staff bioinformaticist to confirm whether EGFR, including exon 19, was studied as part of the July 2020 test.
- Results indicate that EGFR was studied, but in the structural variant analysis pipeline, exon 19 was deemed uncallable.