“We did genetic testing and it came back clear!” Well your VCF file probably says otherwise!

 

In the TikTok/Instagram world where many people live these days, everything is kept very simple. The real world is becoming the alternative reality.

Over the years, many people have sent me their genetic testing results. Nowadays we have moved on to whole genome sequencing (WGS), which generates vast amounts of data and is pretty much as good as it gets. Vast amounts of data brings with it the problem of how to filter through it and not miss something critical.

I have written many times that parents who have had genetic testing carried out on their child should ask for the full list of mutations, not just those on the lab’s list for autism. In the case of whole exome sequencing (WES), this could produce a list of 10 to 30 mutations, which might well include an ion channel, or a similar variant, that is actually relevant to autism.  

More and more parents are doing this or even repeating the WGS elsewhere and then getting a very different interpretation, with a likely causal gene identified.

 

“We did genetic testing and it came back clear!”

I am sorry to disappoint the TikTokers, but nobody’s WGS results can come back clear.

 

What is whole genome sequencing (WGS)?

Scientists have put together a reference genome, based on the results of many different real people. 

In WGS, blood or saliva is used to sequence the entire genome of the patient, and then it is automatically compared to the reference genome. Tens of thousands to millions of variants will be identified. They all end up in the VCF (Variant Call Format) file.

 

The VCF (Variant call format) file

The VCF file is generated by special software. It contains details about the genetic variants.

 

Annotated VCF file

More automation then looks up each of the detected variants to see if they are already listed in databases, as being known to cause problems. If the variant has not been listed in these databases, it will not be highlighted.

These days most labs will provide the annotated VCF file and it can be huge.

 

The role of the geneticist or ChatGPT

The geneticist is then supposed to take the annotated VCF file and filter the results based on the clinical condition of the patient.

Results are categorized as:

·        Pathogenic

·        Likely pathogenic

·        Variant of uncertain significance

·        Likely benign

·        Benign

In theory, the geneticist’s job is to translate the vast and complex information from genetic tests into actionable insights for diagnosis, treatment, and genetic counselling, ensuring both precision and empathy in patient care.

He/she should check whether the identified variant explains the patient's specific symptoms (e.g., developmental delay, speech impairment, intellectual disability).

If multiple variants are identified, consider whether they might interact.

Some individuals may have only one variant, while others may carry several genetic changes that contribute to their condition.

 

How come there are so many apparently sloppy geneticists?

Doctors like to deal with certainties and the geneticist has to avoid diagnosing a gene as causal, when there is a chance it actually is not. They also know that there are almost no genes related to autism that they can treat. There is no incentive a take a stab at causality, when they know they cannot offer any follow-on treatment. Contrast that with the parent's perspective looking for any clues.

Genetics is actually all about probabilities, much more than certainties. So, if you have 3 mutations that are each individually “survivable”, the combination might be causal.

Here is an example:

Imagine a scenario where:

• Mutation 1: Causes a minor disruption in protein folding (survivable).
• Mutation 2: Impairs a metabolic enzyme function (survivable).
• Mutation 3: Affects a signaling pathway that slightly reduces cellular repair efficiency (survivable).

Each of these mutations, on their own, may not lead to a disease. However, when combined, these disruptions might overwhelm cellular systems and cause a disease phenotype (for example, a neurodegenerative condition), as the cumulative effect of these disruptions could impair essential cellular functions beyond a survivable threshold.

  

Not enough people with autism submit WGS data

I would think you need to have the WGS data from at least a million people with an autism diagnosis, from mild to severe, to be pretty sure you have identified the majority of causal mutations. Many of these causal mutations will actually be combinations of a few different genes - it would get very complicated. 

The number of people with autism currently included in the reference databases so far is tiny.  It is biased towards those with profound single-gene autism.

We really need to know about polygenic autism, which accounts for the vast majority of cases. Those cases range from profound to trivial.

I think that if I were in charge, I would tell people that if they want an autism diagnosis, they need to provide a saliva sample. No sample, no diagnosis. The test can be anonymous, if people prefer, so there is nothing to fear. At least this would reduce the waiting time for diagnosis!

  

TikTok autism

On social media you get a lot of self-diagnosed autism, but you also get some Moms/Mums of kids with profound autism.

I saw one today who has 3 sons and 1 daughter, all with profound autism. She says it cannot be genetic, because the geneticist did not find a causal gene.  In theory she might be right. Perhaps there is an environmental explanation, maybe the babies were all exposed to the same toxic environment (food, water, air, a high voltage power cable over the house …)

More likely, the geneticist did not do his job. Or, the kids have mutations that have not yet been added to the list of causal genes. The list grows every day. If she used a good provider, like GeneDX, and she ticked a box on the form, then they may come back to her in 5 years’ time and tell her that one of her kids’ mutations is now recognized as causal. This happened to one reader of this blog.   

My tip would be to ask for the Annotated VCF file and get ChatGPT to analyse it for anything that might explain profound autism.

It looks like any computer-savvy middle-aged person can do this; no prior experience needed!

 

Conclusion

I was pleasantly surprised to hear from several readers recently who followed my suggestion to dig deeper into the results of their child’s genetic testing. They all found something valuable, hidden away. Perhaps you wouldn’t contact Peter if you didn’t find something of use!

To analyze genetic data effectively, you will need the paid version ChatGPT Plus (GPT-4). While it is powerful, it does have processing limits. For example, processing a large VCF file requires breaking it into smaller, manageable parts. Fortunately, ChatGPT can guide you through this process.

If you were doing this as a professional service, you would be better off using GPT-4 via API. This tool, created by the same company, is designed for heavy data processing rather than conversational interaction.

When it comes to genetic forms of autism with approved drug therapies, there seem to be only two: 

·        Tuberous Sclerosis Complex (TSC1 or TSC2 Mutation): Treated with Everolimus (Afinitor) in the US, Canada, EU, UK, and Australia. 

·        Rett Syndrome (MECP2 Mutation): Treated with Trofinetide (Daybue) in the US and Canada if you have the money.

This highlights why geneticists focus on diagnosis rather than treatment. 

Unofficially, the possibilities for treatment are extensive, limited only by creativity and emerging research. Here again, ChatGPT can do much of the work for you.

P.S.  TikTok, Instagram and Facebook can be fun, but for something factual, better stick to ChatGPT or similar services.

P.P.S. I do not have shares in OpenAI, who own ChatGPT