A Conversation with SFARI scientists Alan Packer, Pamela Feliciano and Misia Kowanda about SFARI gene lists

Image of Feliciano Packer Kowanda
From left to right: SPARK director Pamela Feliciano, SFARI senior scientist Alan Packer, and former Simons Searchlight outreach manager Misia Kowanda.

Genetic studies of autism spectrum disorder (ASD) have identified hundreds of genes associated with the condition. As this knowledge accumulates, a systematic classification of ASD risk genes can help inform new research studies and provide clinical information to families about the causes of their autism. To that end, the Simons Foundation Autism Research Initiative (SFARI) maintains SFARI Gene, a database listing over 1,000 ASD genes that are ranked according to their strength of association with the condition, and two additional lists of genes associated with ASD and related conditions that are curated, respectively, by SPARK and Simons Searchlight. The SPARK gene list currently includes 167 single genes, 43 copy number variants and 5 chromosomal differences linked to autism. The Simons Searchlight gene list encompasses 152 single genes and 23 copy number variants linked to autism and other genetic neurodevelopmental conditions.

I recently spoke with Alan Packer, senior scientist at SFARI, Pamela Feliciano; senior scientist at SFARI and SPARK director; and Misia Kowanda, former outreach manager at Simons Searchlight, to learn more about each of SFARI’s gene lists and the ways they can facilitate the understanding of and research on ASD genetics.

The interview has been edited for clarity and brevity.

SFARI maintains three lists of genes that are associated with ASD, one for SFARI Gene, one for SPARK, and one for Simons Searchlight. Why were these separate lists created?

Alan Packer: When I first joined SFARI in 2009, there was an excellent early version of SFARI Gene that listed all genes that have some link with autism. The problem in my mind was that there was no information as to how robust the evidence was for each gene’s association with ASD, whether it was strong or weak. I thought there was an opportunity to work with advisors and geneticists in the community who could help us come up with some guidelines for determining what the strength of the evidence was for each gene and then use that knowledge to develop a comprehensive list of genes with at least some evidence that they’re involved in autism risk. That list evolved over time, and eventually, we subdivided the genes into seven different categories based on the level of evidence for their role in ASD risk. More recently, these categories were redefined and condensed into four:  S — syndromic, category 1 — high confidence, category 2 —strong candidate, and category 3 — suggestive evidence. Today, there are over a thousand genes on the list.

Pamela Feliciano: Like SFARI Gene, the SPARK gene list came out of a need for a list of ASD risk genes that met a high bar of genetic evidence. It was largely about defining what types of genetic variants we would return in SPARK as well as providing a curated gene list to help researchers understand what genes we think have the highest levels of evidence for autism. At first, most were SFARI Gene category 1 genes, but there were some historical genes that were missing from that list. The most recent version of the SPARK gene list has 167 single genes, 43 different copy number variants and 5 chromosomal aneuploidies.

Misia Kowanda: The Simons Searchlight gene list is derived from the SPARK gene list, but it doesn’t include all the same genes. Right now, we have 152 single genes, 23 copy number variants and no chromosomal conditions such as trisomies or monosomies. There are a few additional factors that are important for Simons Searchlight specifically. We include genes from the SPARK gene list that don’t have an already existing genetic registry and conditions that have a complementary registry. We also have some genes that SPARK doesn’t include because they haven’t had a returnable result in SPARK as of yet [editorial note: SPARK has returned genetic results to more than 500 study participants to date].

“The SPARK gene list is primarily used by researchers wanting to understand what we're returning to participants.”

Pamela Feliciano

How are these lists used?

Alan Packer: SFARI Gene is primarily used by the basic research community, but not exclusively. We try to make the site as comprehensive as possible. There are a number of genes we list with only modest evidence that implicates the gene in autism. Unlike SPARK or Simons Searchlight, these aren’t genes that would ever be returned to families, but they would still be of interest to basic researchers who want to better understand the total landscape of genes that might be involved in autism.

I think SFARI Gene is used by researchers in two general ways. One is that if a researcher has a particular interest in gene X, they can go to the site and search for that gene and find out all the information that might implicate that gene in autism, including human genetic evidence and data from animal models. There are other people like systems biologists who use SFARI Gene to see if there are particular gene networks that might be interesting in this list of a thousand genes and try to see what they can make out of that. I think the database gets some use in both of those areas.

Pamela Feliciano: The SPARK gene list is primarily used by researchers wanting to understand what we’re returning to participants. Their research might be more translationally or clinically oriented, but it ultimately comes down to helping researchers judge whether a gene is strongly implicated in autism.

Misia Kowanda: The Simons Searchlight list is used as an eligibility list for enrollment, therefore it is used by individuals and families, and genetics professionals such as genetic counselors and clinicians so that they know to refer people to Simons Searchlight. It’s also used by researchers who are studying our cohorts, both clinical and basic researchers. We also publicly list the genetic variants found in our cohort and available biospecimens.

“The system is designed so that a researcher can come to the SFARI Gene site, see a gene on the list and find all of the supporting evidence for that gene somewhere in the published literature and look at the evidence for themselves. Essentially, it's a take-all-comers approach. ”

Alan Packer

You’ve mentioned that there have been differences between SFARI Gene and the SPARK gene list in the past. What are the criteria used to determine inclusion in each list, and are there any criteria that would exclude certain genes from being included in certain lists?

Alan Packer: SFARI Gene includes any gene for which there’s any evidence at all implicating it in autism in the literature. Obviously, that’s why there are so many genes on the list. We do have a category 1 list that only includes the genes in autism where we’ve decided that the evidence for an association with ASD is rock solid and that it would make sense to have it follow the SPARK list, which has a high standard of evidence. For this top category, SFARI Gene now closely follows SPARK, although it does also include the genes from the 2020 paper by Satterstrom and colleagues1. For the genes with lower levels of evidence, we do our own ranking.

One critical thing to point out is that we rely entirely on the published literature. If it’s not in PubMed, for example, it won’t go into SFARI Gene. The system is designed so that a researcher can come to the SFARI Gene site, see a gene on the list and find all of the supporting evidence for that gene somewhere in the published literature and look at the evidence for themselves. Essentially, it’s a take-all-comers approach.

“If we find that more than 20 percent of people with the condition have also been diagnosed with autism, we’ll consider it for [the Simons Searchlight] list.”

Misia Kowanda

Pamela Feliciano: As Alan mentioned, genes on the SPARK gene list need to meet a high level of genetic evidence. What we want researchers to understand is that we are focused on returning variants in genes that have a clear role in autism. There are a few exceptions where we have added genes to the SPARK list because the syndrome associated with those genes often includes autism. An example of this is Noonan syndrome; people with Noonan syndrome have a high incidence of autism. Because of that, we have added most of the genes that cause Noonan syndrome to the SPARK gene list.

Misia Kowanda: If we can find about 20 percent or more of people with the condition have also been diagnosed with autism, we’ll consider it for our list.

“These lists have evolved over time to serve different purposes. Over the years, we’ve gone back and looked at the lists and modified them to better serve the community’s needs.”

Pamela Feliciano

How much overlap exists between the lists? What are the significant differences?

Alan Packer: The SPARK gene list significantly overlaps with the top category of SFARI Gene. Pam mentioned that genes in SPARK show evidence of a clear loss of function mutation in four individuals in a cohort; we wouldn’t require that for the lower categories in SFARI Gene because there might be evidence in the literature for, say, one or two mutations in that gene which would not be statistically significant enough to say that it’s a true autism risk gene, but we feel like it’s important for the community to be aware of it because it might become a bona fide risk gene.

Pamela Feliciano: The SPARK gene list is highly overlapping with SFARI Gene, but as Alan said, there are many genes we don’t include in the SPARK list because the evidence isn’t quite conclusive. For the SPARK and Simons Searchlight lists, most of the genes that are on SPARK’s but not Simons Searchlight’s are genes where the community already has a registry.

Misia Kowanda: For Simons Searchlight, in terms of differences from SPARK, I’d add that for us it sometimes makes scientific sense to include groups of conditions. For example, we were more inclusive with GRIN genes because even though diagnoses of autism may be lower within the community, it scientifically makes sense to group common pathways or a common function, so this justifies including them. We want to be a resource to the research community and help bring similar communities together.

Venn diagram of where genes overlap between initiatives
Shared genes. An illustration of the relationship between the gene lists for SPARK and Simons Searchlight, based on the current version of the lists (SPARK gene list, updated September 2021; Simons Searchlight gene list, updated March 2021). The SPARK gene list has 167 genes, 35 of which are not found on the Simons Searchlight gene list. The Simons Searchlight gene list has 152 genes, 20 of which are not found on the SPARK gene list. The two share 132 genes altogether.

How often are the lists updated?

Alan Packer: SFARI Gene is typically updated quarterly with curated data from the literature that are provided by the team at MindSpec. For the category 1 genes, whenever the SPARK list is updated, we would just import any additions into SFARI Gene.

Pamela Feliciano: We update the SPARK list multiple times a year.

Misia Kowanda: We tend to look at the Simons Searchlight list when SPARK has their quarterly medical genetic meeting, which is when they usually review their list.

Any final comments?

Alan Packer: The one thing I’d add is that even though we categorize genes in SFARI Gene in a certain way, it’s not the only reasonable way to rank these genes. We try to provide links for each gene to other scoring efforts in the community and to categorize autism risk genes so people can see what the range of possibilities is and then decide for themselves which one they think is the most helpful. Having three separate lists helps because they all complement each other in illustrative ways.

Pamela Feliciano: And I should say that these lists have evolved over time to serve different purposes. Over the years, we’ve gone back and looked at the lists and modified them to better serve the community’s needs.

Reference

  1. Satterstrom F.K. et al., Cell 180, 568-584 (2020) PubMed
Recent SFARI Conversations