parkmoll commited on
Commit
1249397
·
verified ·
1 Parent(s): 843720e

Upload 2 files

Browse files

Push updated rubrics

Files changed (2) hide show
  1. question_types.json +15 -53
  2. rare_variants_qa.json +1 -1
question_types.json CHANGED
@@ -2,17 +2,13 @@
2
  "pathomechanism": {
3
  "template": "Context: Your task is to determine the pathomechanism of a given pathogenic gene variant. A pathogenic variant can lead to different effects. The variant can lead to a loss of function (LoF) of a protein, a toxic gain of function (GoF), or a dominant-negative (DN) effect. Assessing the variant effect can be done by by retrieving publications and reports of the variant. Though if no functional studies from these resources are found, use the provided considerations:\n(1) Variants with a predicted loss-of-function (LoF) effect, like nonsense and frameshift variants, can be assumed in many clinical contexts to result in a null allele in the absence of functional studies. However, nonsense or frameshift variants leading to a premature stop codon in the last exon or within 50 bp of 3' end of the penultimate exons may not necessarily lead to loss-of-function. In this case, you should take into consideration protein domains and presence of downstream pathogenic variants. In a disease that has only been associated with LoF variants, a newly reported likely pathogenic or pathogenic missense variant may be LoF, especially if this variant is in trans with a known pathogenic LoF variant (for recessive diseases).\n(2) For genes where both LoF and GoF are a known cause of disease, assessing the pathomechanism of a missense variant is challenging. Here, it will be important to take the phenotype and inheritance pattern of the variant into account to make a decision on the pathomechanism of the variant. It can be possible to predict a pathomechanism in cases where LoF and GoF variants lead to distinguishable phenotypes, though there are only rare cases where phenotypes are distinct enough to make such a decision. When in doubt, address that more evidence is necessary and you should suggest what functional studies would be needed to confirm.\n\nQuestion: What is the pathomechanism of variant {variant}?",
4
  "rubrics": [
5
- {
6
- "name": "citation",
7
- "criterion": "All citation worthy statements are cited. These refer to statements that reference findings from retrieved resources (e.g. search results, webpages, retrieved papers, etc.)."
8
- },
9
  {
10
  "name": "evidence_quality",
11
- "criterion": "Confirms the claimed pathomechanism with a functional assay, or addresses that sufficient functional evidence is not available. Assign half credit if evidence is present, but not a functional assay or related to a different variant in the same gene."
12
  },
13
  {
14
  "name": "consistency",
15
- "criterion": "Statements with citations are consistent with their corresponding cited text. If available, the original text for each citation id will be put in the 'Cited texts' section."
16
  },
17
  {
18
  "name": "synthesis",
@@ -24,38 +20,13 @@
24
  "splicing_effects": {
25
  "template": "Context: Your task is to research the splicing effects of a given gene variant. A variant's impact on splicing (or lack thereof) must be supported by functional assay, i.e. RNAseq, qPCR, or cDNA sequencing/analysis obtained from patient-derived cells. Data gathered through mini- and midi-genes cannot be considered sufficient as these assays do not take the full genetic environmental context into account and results can be misleading. If splicing effects are confirmed, the exact effects on splicing need to be evaluated. This can be a gain of an acceptor or donor splice site or the loss/weakening of an acceptor or donor splice site. \n\nQuestion: How does the genetic variant {variant} impact splicing?",
26
  "rubrics": [
27
- {
28
- "name": "citation",
29
- "criterion": "All citation worthy statements are cited. These refer to statements that reference findings from retrieved resources (e.g. search results, webpages, retrieved papers, etc.)."
30
- },
31
  {
32
  "name": "evidence_quality",
33
- "criterion": "Confirms reported splicing effect (or lack thereof) with a functional assay (RNAseq, qPCR, or cDNA sequencing/analysis from patient-derived cells) from the literature, or addresses that no functional evidence is available. If the response claims the presence or absence of a splicing effect, assign full points only if the claim is supported by a cited functional assay. If the response addresses that no functional evidence is available and voices uncertainty about the splicing effect, still assign full credit. Assign half credit if evidence is present, but not a functional assay or related to a different variant in the same gene."
34
  },
35
  {
36
  "name": "consistency",
37
- "criterion": "Statements with citations are consistent with their corresponding cited text. The original text for each citation id will be put in the 'Cited texts' section."
38
- },
39
- {
40
- "name": "synthesis",
41
- "criterion": "Response cites multiple sources, and has at least one statement discussing how sources support or conflict with each other. This might manifest as a single statement with multiple supporting citations. Award no points if the response only considers a single source."
42
- }
43
- ]
44
- },
45
- "functional_domain": {
46
- "template": "Context: Your task is to analyze the functional domains affected by a given gene variant. Please check the exact role of a domain and the effect of losing the domain. Resources like UniProt, the Protein Data Bank, or Pubmed can help with this. Pay attention to the role of domains overlapping the variant (i.e., DNA binding domain, catalytic domains) or amino acids (i.e., specific amino acid known to play an important role in enzyme activity or protein structure). A web search for the protein name and structure can also be helpful.\n\nQuestion: Provide an overview of the functional domains impacted by the variant {variant}.",
47
- "rubrics": [
48
- {
49
- "name": "citation",
50
- "criterion": "All citation worthy statements are cited. These refer to statements that reference findings from retrieved resources (e.g. search results, webpages, retrieved papers, etc.)."
51
- },
52
- {
53
- "name": "intermediate_step",
54
- "criterion": "Response uses evidence to determine the functional domain(s) that the exon of the variant encodes. Assign full points if present, no points if not."
55
- },
56
- {
57
- "name": "consistency",
58
- "criterion": "Statements with citations are consistent with their corresponding cited text. The original text for each citation id will be put in the 'Cited texts' section."
59
  },
60
  {
61
  "name": "synthesis",
@@ -67,12 +38,12 @@
67
  "template": "Context: Reports on inheritance patterns of diseases associated with the gene variant will be helpful here. Some genes are implicated in different diseases which can have different inheritance patterns. In these cases, identify which pattern of inheritance applies to the specific case. The variant may have been reported in a publication where a pattern of inheritance is noted. The gnomAD population frequency of a variant can also indicate if the variant is associated with a dominant or recessive inheritance. For example, if a LoF variant has a high allele frequency for heterozygotes but no homozygotes are reported, the variant is more likely to be associated with an AR pattern of inheritance. Similarly, for X-linked disorders in XY males, a lack of hemizygotes would be the equivalent assessment.\n\nQuestion: What is the inheritance pattern of diseases associated with variant {variant}?",
68
  "rubrics": [
69
  {
70
- "name": "citation",
71
- "criterion": "All citation worthy statements are cited. These refer to statements that reference findings from retrieved resources (e.g. search results, webpages, retrieved papers, etc.)."
72
  },
73
  {
74
  "name": "consistency",
75
- "criterion": "Statements with citations are consistent with their corresponding cited text. The original text for each citation id will be put in the 'Cited texts' section."
76
  },
77
  {
78
  "name": "synthesis",
@@ -84,21 +55,16 @@
84
  "template": "Context: Your task is to determine the haploinsufficiency status of a given disease-causing variant. Haploinsufficiency refers to a situation in which one healthy, wildtype allele does not generate sufficient protein product to preserve the physiological state. In most cases, haploinsufficiency is connected to autosomal dominant disorders associated with LoF variants; however, several genes have been identified that are associated with LoF variants causing haploinsufficiency, and also GoF variants. A gene-level dosage sensitivity curation from the ClinGen consortium is considered the highest level of evidence for haploinsufficiency. pLI and LOEUF scores in gnomAD can also be used, though they are not sufficient without functional evidence from the literature. Further reports can be found in the literature through OMIM, PubMed, GeneReviews, and OrphaNet.\n\nQuestion: Is variant {variant} associated with haploinsufficiency?",
85
  "rubrics": [
86
  {
87
- "name": "citation",
88
- "criterion": "All citation worthy statements are cited. These refer to statements that reference findings from retrieved resources (e.g. search results, webpages, retrieved papers, etc.)."
89
  },
90
  {
91
  "name": "consistency",
92
- "criterion": "Statements with citations are consistent with their corresponding cited text. The original text for each citation id will be put in the 'Cited texts' section."
93
- }
94
- ]
95
- },
96
- "exon_analysis": {
97
- "template": "Context: When determining exon positions and sizes, use GENCODE annotations over NCBI RefSeq for consistency. GENCODE annotations can be accessed using the UCSC Genome Browser. For negative strands, exons should be numbered from highest to lowest genomic coordinate. Otherwise, number exons from lowest to highest genomic coordinates. \n\nQuestion: What exon is variant {variant} on, and what proportion of the coding region does it take up? Calculate the proportion as (number of codons in the exon) / (number of total amino acids in the protein).",
98
- "rubrics": [
99
  {
100
- "name": "citation",
101
- "criterion": "All citation worthy statements are cited. These refer to statements that reference findings from retrieved resources (e.g. search results, webpages, retrieved papers, etc.)."
102
  }
103
  ]
104
  },
@@ -107,15 +73,11 @@
107
  "rubrics": [
108
  {
109
  "name": "evidence_quality",
110
- "criterion": "If a study on an existing therapy is found, the response must either mention functional evidence for the therapy's safety/efficacy or the lack thereof."
111
- },
112
- {
113
- "name": "citation",
114
- "criterion": "All citation worthy statements are cited. These refer to statements that reference findings from retrieved resources (e.g. search results, webpages, retrieved papers, etc.)."
115
  },
116
  {
117
  "name": "consistency",
118
- "criterion": "Statements with citations are consistent with their corresponding cited text. The original text for each citation id will be put in the 'Cited texts' section."
119
  },
120
  {
121
  "name": "synthesis",
 
2
  "pathomechanism": {
3
  "template": "Context: Your task is to determine the pathomechanism of a given pathogenic gene variant. A pathogenic variant can lead to different effects. The variant can lead to a loss of function (LoF) of a protein, a toxic gain of function (GoF), or a dominant-negative (DN) effect. Assessing the variant effect can be done by by retrieving publications and reports of the variant. Though if no functional studies from these resources are found, use the provided considerations:\n(1) Variants with a predicted loss-of-function (LoF) effect, like nonsense and frameshift variants, can be assumed in many clinical contexts to result in a null allele in the absence of functional studies. However, nonsense or frameshift variants leading to a premature stop codon in the last exon or within 50 bp of 3' end of the penultimate exons may not necessarily lead to loss-of-function. In this case, you should take into consideration protein domains and presence of downstream pathogenic variants. In a disease that has only been associated with LoF variants, a newly reported likely pathogenic or pathogenic missense variant may be LoF, especially if this variant is in trans with a known pathogenic LoF variant (for recessive diseases).\n(2) For genes where both LoF and GoF are a known cause of disease, assessing the pathomechanism of a missense variant is challenging. Here, it will be important to take the phenotype and inheritance pattern of the variant into account to make a decision on the pathomechanism of the variant. It can be possible to predict a pathomechanism in cases where LoF and GoF variants lead to distinguishable phenotypes, though there are only rare cases where phenotypes are distinct enough to make such a decision. When in doubt, address that more evidence is necessary and you should suggest what functional studies would be needed to confirm.\n\nQuestion: What is the pathomechanism of variant {variant}?",
4
  "rubrics": [
 
 
 
 
5
  {
6
  "name": "evidence_quality",
7
+ "criterion": "Confirms the claimed pathomechanism with at least one functional study, or addresses that sufficient functional evidence for the variant is not available. Evidence must have a citation and be related to the given variant in order to received credit for this criterion."
8
  },
9
  {
10
  "name": "consistency",
11
+ "criterion": "Statements with citations are consistent with their corresponding cited text. This score for this criterion should reflect the proportion of cited statements that are consistent with their source text. If available, the original text for each citation id will be put in the 'Cited texts' section."
12
  },
13
  {
14
  "name": "synthesis",
 
20
  "splicing_effects": {
21
  "template": "Context: Your task is to research the splicing effects of a given gene variant. A variant's impact on splicing (or lack thereof) must be supported by functional assay, i.e. RNAseq, qPCR, or cDNA sequencing/analysis obtained from patient-derived cells. Data gathered through mini- and midi-genes cannot be considered sufficient as these assays do not take the full genetic environmental context into account and results can be misleading. If splicing effects are confirmed, the exact effects on splicing need to be evaluated. This can be a gain of an acceptor or donor splice site or the loss/weakening of an acceptor or donor splice site. \n\nQuestion: How does the genetic variant {variant} impact splicing?",
22
  "rubrics": [
 
 
 
 
23
  {
24
  "name": "evidence_quality",
25
+ "criterion": "Confirms reported splicing effect (or lack thereof) with a functional assay (RNAseq, qPCR, or cDNA sequencing/analysis from patient-derived cells) cited from a research paper. Evidence must have a citation and be related to the given variant in order to received credit for this criterion."
26
  },
27
  {
28
  "name": "consistency",
29
+ "criterion": "Statements with citations are consistent with their corresponding cited text. This score for this criterion should reflect the proportion of cited statements that are consistent with their source text. The original text for each citation id will be put in the 'Cited texts' section."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30
  },
31
  {
32
  "name": "synthesis",
 
38
  "template": "Context: Reports on inheritance patterns of diseases associated with the gene variant will be helpful here. Some genes are implicated in different diseases which can have different inheritance patterns. In these cases, identify which pattern of inheritance applies to the specific case. The variant may have been reported in a publication where a pattern of inheritance is noted. The gnomAD population frequency of a variant can also indicate if the variant is associated with a dominant or recessive inheritance. For example, if a LoF variant has a high allele frequency for heterozygotes but no homozygotes are reported, the variant is more likely to be associated with an AR pattern of inheritance. Similarly, for X-linked disorders in XY males, a lack of hemizygotes would be the equivalent assessment.\n\nQuestion: What is the inheritance pattern of diseases associated with variant {variant}?",
39
  "rubrics": [
40
  {
41
+ "name": "evidence_quality",
42
+ "criterion": "The response must cite at least one publication/report where the pattern of inheritance for the variant or its associated disease is reported. Evidence must have a citation and be related to the given variant in order to received credit for this criterion."
43
  },
44
  {
45
  "name": "consistency",
46
+ "criterion": "Statements with citations are consistent with their corresponding cited text. This score for this criterion should reflect the proportion of cited statements that are consistent with their source text. The original text for each citation id will be put in the 'Cited texts' section."
47
  },
48
  {
49
  "name": "synthesis",
 
55
  "template": "Context: Your task is to determine the haploinsufficiency status of a given disease-causing variant. Haploinsufficiency refers to a situation in which one healthy, wildtype allele does not generate sufficient protein product to preserve the physiological state. In most cases, haploinsufficiency is connected to autosomal dominant disorders associated with LoF variants; however, several genes have been identified that are associated with LoF variants causing haploinsufficiency, and also GoF variants. A gene-level dosage sensitivity curation from the ClinGen consortium is considered the highest level of evidence for haploinsufficiency. pLI and LOEUF scores in gnomAD can also be used, though they are not sufficient without functional evidence from the literature. Further reports can be found in the literature through OMIM, PubMed, GeneReviews, and OrphaNet.\n\nQuestion: Is variant {variant} associated with haploinsufficiency?",
56
  "rubrics": [
57
  {
58
+ "name": "evidence_quality",
59
+ "criterion": "The response must cite at least one publication/report where the haploinsufficiency status of the variant is reported. Reports from any of the evidence sources mentioned in the context are also considered valid. A gene-level dosage sensitivity curation from the ClinGen consortium is also considered valid for full credit. Evidence must have a citation and be related to the given variant in order to received credit for this criterion."
60
  },
61
  {
62
  "name": "consistency",
63
+ "criterion": "Statements with citations are consistent with their corresponding cited text. This score for this criterion should reflect the proportion of cited statements that are consistent with their source text. The original text for each citation id will be put in the 'Cited texts' section."
64
+ },
 
 
 
 
 
65
  {
66
+ "name": "synthesis",
67
+ "criterion": "Response cites multiple sources, and has at least one statement discussing how sources support or conflict with each other. This might manifest as a single statement with multiple supporting citations. Award no points if the response only considers a single source."
68
  }
69
  ]
70
  },
 
73
  "rubrics": [
74
  {
75
  "name": "evidence_quality",
76
+ "criterion": "If a study on an existing therapy is found, the response must either mention functional evidence for the therapy's safety/efficacy or the lack thereof. Evidence must have a citation and be related to the given variant in order to received credit for this criterion. For exon skipping therapies, evidence is valid for full credit as long as it is for the exon that the given variant is located in (even if it doesn't mention the same exact variant)."
 
 
 
 
77
  },
78
  {
79
  "name": "consistency",
80
+ "criterion": "Statements with citations are consistent with their corresponding cited text. This score for this criterion should reflect the proportion of cited statements that are consistent with their source text. The original text for each citation id will be put in the 'Cited texts' section."
81
  },
82
  {
83
  "name": "synthesis",
rare_variants_qa.json CHANGED
@@ -366,7 +366,7 @@
366
  "rubrics": [
367
  {
368
  "name": "correct_answer",
369
- "criterion": "Response mentions variant results in alternatate splicing."
370
  },
371
  {
372
  "name": "correct_answer_detail",
 
366
  "rubrics": [
367
  {
368
  "name": "correct_answer",
369
+ "criterion": "Response mentions variant results in alternate splicing."
370
  },
371
  {
372
  "name": "correct_answer_detail",