Curate new ASC and fat distribution burden evidence

[ireneisdoomed]

The results of burden tests identifying 60 genes associated with autism spectrum disorder have recently been published (PMID:35982159).

The analysis has been done primarily on the SPARK cohort of approximately 40k individuals, making it the largest study to date, and we would like to bring these results to the platform by adding them to our 'curation' spreadsheet.

Data availability

These are 5 novel targets and some metadata about the analysis.

The full statistics for the rest of the genes are available in the Supp Table 9.

[ireneisdoomed]

This curation has been done and is ready to review in PR#11.

Some personal notes about the autism publication:

Article: https://www.nature.com/articles/s41588-022-01148-2

Note: the study reports 60 but when looking at the data (Supp Table 9) only 56 are found significant. We are capturing all of the novel genes.

New risk genes for autism were identified by analysing de novo and rare inherited variants from WES and WGS data

• 60 genes
  • 5 novel: NAV3, MARK2, SCAF1, ITSN1, HNRNPUL2
  • Threshold of significance: P < 2.5 × 10−6 (exome-wide)
  • 42,607 individuals (35,130 coming from the SPARK cohort: a study aimed at collecting data from a large numbers of individuals with autism across the spectrum and their available family members)
  • The study is cross ancestry: SPARK collects data from American people and other cohorts include european and african individuals, but the analysis was not ancestry based.

Genetic burden is analysed in 2 stages:

1. Proper gene burden analysis. 3 strategies:

   1. Assessment of DNVs in ASD trios
   • Characterization of the enrichment pattern in known and candidate risk genes
   • Analysis of mutation intolerance (probability of being LoF intolerant by using ExAC pLI and gnomAD)
   • Gene-based burden tests of LoF and missense DNVs by **DeNovoWEST 5.
   2. Assesment of transmission of rare LoF variants from parents without ASD diagnoses or intellectual disability to offspring with ASD
   • Estimation of the overtransmission from unaffected parents to ASD offspring in all genes and gene sets predefined by functional genomic data or results from DNV analysis
   3. Analysis of increased rate of LoFs in cases vs population controls

2. Meta analysis of the results of both strategies:

   • The samples include all data from stage1 and stage2:
     • 6,174 new ASD trios
     • 1,942 new duos
     • 15,780 unrelated cases
     • 236,000 population controls.
   • Only high confidence ultra rare LoF variants (MAF < 5E-5) were considered

In the article it is described that intellectual disability hasn't been observed for these novel genes, compared with other well-known ASD risk genes like CHD8. At the same time, the data show that carriers of variants in the novel genes contribute to other neuro conditions.
This is a sign that these risk genes with a modest effect size may represent a different class of ASD genes that are more directly associated with core symptoms of ASD and/or neuropsychiatric conditions rather than global brain development and intellectual disability.

[ireneisdoomed]

PR#11 also includes the results of a publication that describes the protective effect on some variants on fat distribution. The publication also has gene level results.

Some personal notes about the fat distribution publication
16 genes associated with fat distribution were identified by analysing missense variants from WES data. Most of them with a protective direction of effect.
- 16 genes
- 12 novel, 4 previously known (ACVR1C, CALCRL, PLIN1, PDE3B)
- INHBE is particularly interesting because of its favorable fat distribution, favorable metabolic profile and protection from type 2 diabetes for heterozygous PTVs
- Threshold of significance: P < 3.6 × 10−7 (exome-wide)
- Primary trait of interest: BMI-adjusted WHR
- They also describe evidence of sex-dimorphism
- Associations have consistent effect estimates across ancestries

The cohort consists of 618375 individuals from UK Biobank, MDCS (Swedish), MCPS (Mexican). Most of them come from UKB.
In the Supp Table 3, the gene burden analyses are disclosed per ancestry and per cohort, unlike the Table 1 where that data is aggregated. We would be losing some associations if we were to use the disclosed results, since the statistical power is lower (for example, CD36 would be underpowered). After discussion we have decided to report the aggregated results.