View Full Version : Dante WGS and Promethease?

12-23-2018, 08:43 AM
Does anybody know how to prepare gVCF file suitable for Promethease service? It looks like the delivered VCF file is not suitable because it lacks the genotypes that completely match the reference in question.

Using the delivered BAM file should be possible to create the best suitable gVCF file, does anybody know how to do it either by using sequencing.com or by the linux tools directly?

01-04-2019, 12:39 AM
I found that the best way is to upload the BAM file to Promethease.

06-10-2019, 06:50 AM
May i ask which steps did you take to upload your Dante Labs BAM file to Promethease ? I am getting errors uploading it

06-10-2019, 06:34 PM
IIRC I had the BAM file in my Dropbox and I entered the download link.

06-11-2019, 01:00 AM
IIRC I had the BAM file in my Dropbox and I entered the download link.

By any chance do you know what's the best alignment for Promethease use?

06-11-2019, 11:54 AM
By any chance do you know what's the best alignment for Promethease use?

I've been told that Hg38 has more medically relevant SNPs than older Hg37/hg19. Though Promethease has removed lots of medically relevant information from their site due to FDA regulations.. so it's not so complete anymore.

06-11-2019, 12:35 PM
Is there reference for FDA snipping Promethease? I know that's been long speculated, but I've not seen any mention of it on any forums, SNPedia or the analysis yet. Due to the nature of Promethease/SNPedia, if the FDA rules were to apply to it, I don't think it could operate at all as there is no way the reports can be validated, tested for user comprehension, and almost all of the pages deal with un-actionable/un-certain variants.
By the way, Promethease relies entirely on volunteer-written summaries from SNPedia, so it's never been anywhere near "complete" (except for the automatically aggregated data from GWAS catalog, allele frequency, gene names etc.)
I don't think the reference genome makes so large difference in terms of medically relevant variants, especially since many places still use the older reference, but in principle it reduces genotypin errors (Assuming everything's done correctly). Now I'm curious if one can easily pull a list of SNPedia variants which don't exist on older build... The reverse sometimes happens: https://www.snpedia.com/index.php/Rs2519093

06-11-2019, 02:20 PM
Last checkbox on promethease.com:

I understand that my Promethease report will not include information about predicted response to any specific medications.

According to this reddit post (https://www.reddit.com/r/promethease/comments/b8ofu1/promethease_said_i_had_this_shit_gene_rs1800497tt/ejz3j63/) they think the FDA regulations apply to them:

A decision was made to remove medication-related information from Promethease reports, in accordance with a published FDA memo about the use of such information in direct-to consumer genomic data.

Ultimately I hope someone appeals it, because this should be mostly about presentation. If any service that links known mutations to medical papers with unverified correlations is considered illegal, they'd have to shut down dbSNP and ClinVar.
IMO, as long as you put in a fat disclaimer, stressing that this is raw research and not actionable advice, they have to allow it.
I get their reasoning that some consumers might increase dosages on their own etc, but as long as alternative medicine companies are allowed to peddle their stuff, heavily implying it beats chemotherapy and hiding behind a small disclaimer, I can't see how they can clamp down on what is essentially a research database with an aggregation feature.

On the other hand, Dante Labs itself has almost no disclaimers before you see the reports itself, and they should be forced to be much more upfront about the accuracy of what they are selling.

06-11-2019, 04:29 PM
Ah, right, I forgot the pharmacogenetic tests, I did see those mentioned. However, note that FDA regulates Carrier Screening Tests, Genetic Health Risks Tests, Cancer Predisposition Tests and Pharmacogenetics Tests. Promethease has reports on all those categories (one might argue primarily only those categories) and does not meet FDA's requirements for any of them: https://www.fda.gov/medical-devices/vitro-diagnostics/direct-consumer-tests

I shouldn't really comment, but FDA has implied very heavily between the lines that pharmacogenetic reports are allowed, as long as they do not mention (American) brand-name drugs. Because "Drug X may be dangerous to you" would be bad for marketing, and adverse reactions or even death aren't a public matter. I can sympathize with the other concerns, like Joe Schmock writes on SNPedia that according to 10 patient study drug A may cause nausea, and consumers go telling their doctors they won't take any drug beginning with the letter A... But frankly that will happen with or without genomics, and allowing reports using generic product names only makes the confusion worse. Also you'd be allowed to report the FDA allowed pharmacogenomic interactions: https://www.fda.gov/drugs/science-research-drugs/table-pharmacogenomic-biomarkers-drug-labeling which is really awkward for a DTC report. And no research, please.

FDA: "The FDA has issued a safety communication to alert the public of concerns regarding pharmacogenetic tests with unapproved claims to predict an individual's response to a specific therapeutic drug where these claims may not supported by clinical evidence." (Emphasis mine)

It's hard to drive the line between "medical advice" and "literature retrieval system" that Promethease purpots to be. The problem becomes when people on social media are trying to press Promethease as ultimate truth-sayer on all things genetics instead of the incomplete collaborative research platform it is.

The actual Dante Labs pharmacogenomics report leads in with the following (That is different pre-ample from the sample report on the site, although their Terms of Service includes even stronger disclaimers. 23andMe showed that no amount of disclaimers is enough for FDA though):


Dante Labs Pharmacogenetics Report is for informational purposes only.

The Pharmacogenetics Report analyzes a large amount of genomic data, associating genetic variants found in the genomic files with variants known from the scientific literature. While this Report does not require FDA/EMA approval, we do want to point out that it has not been approved by the FDA/EMA for such use.

The genetic analysis and reporting are based on information from one or more published third party scientific and medical studies. We do not independently judge the validity or accuracy of such published scientific information.

Because scientific and medical information changes over time, your risk assessment and genetically tailored prevention for one or more of the medications contained within this report may also change over time.

Therefore, this report may not be 100% accurate (e.g., new research could mean different results) and may not predict actual results or outcomes. A personís risk of any particular phenotype, condition or trait is also based on other factors not yet analyzed in this report (e.g., diet, lifestyle, etc.).

This report may be updated from time to time so that the analysis and reporting incorporates new or changed research or scientific results. Because of this, the reports produced may change over time.

This genetic report should not be used in place of a visit with or advice from your doctor or other qualified healthcare professional. You should always get the advice of your doctor or other appropriate health care professional if you have any question about diagnosis, treatment, prevention, mitigation, or cure of any medical condition, phenotype, condition, impairment, or the status of your health. Do not stop any medications you have been prescribed, start any new medications, or modify any medical treatments ordered by your healthcare provider without first talking with your healthcare provider. If you have any healthcare related questions, please promptly consult your physician or other qualified healthcare provider.

EDIT: Corrected line-breaks for readability.

06-12-2019, 09:55 AM
IIRC I had the BAM file in my Dropbox and I entered the download link.

Well, that did not work out for me unfortunately. Dropbox seems to have an upper filesize limit of 50Gb. My BAM File exceeds this - its 98GB. Did you compress or split your file perhaps ? Or was it under 50GB ?

06-12-2019, 07:13 PM
I shouldn't really comment, but FDA has implied very heavily between the lines that pharmacogenetic reports are allowed, as long as they do not mention (American) brand-name drugs.So reporting about acetylsalicylic acid would be fine, as long as they don't mention Aspirin? I thought that was what Promethease was doing anyway?

Yes the disclaimers in the report and in the ToS are thorough, but I think this whole notion of "it's research with error bars, not facts and health advice" needs to be featured directly on the product page. Right now, they advertise:

Report on 100+ common diseases, common hereditary cancers, 30+ wellness and longevity traits
To me, they are implying that you get screening for hereditary cancers with yes/no answers, and that is specifically what the FDA wants to (and should) regulate and mandate minimum standards of sensitivity/specificity for.
It's very rare that companies get away with selling you X while putting (well actually, not really X) into page 27 of their terms of service.

06-13-2019, 12:20 AM
I'm not a lawyer, so of course this is mostly personal interpretation. On the main page at https://www.fda.gov/medical-devices/safety-communications/fda-warns-against-use-many-genetic-tests-unapproved-claims-predict-patient-response-specific#actions FDA says, "Most firms addressed the FDA’s concerns by removing specific medication names from their labeling, including promotional material and patient test reports." Feel free to interpret "specific medication names" how you will. It's important to note though that there's a difference between what is decreed and prohibited, and what's enforced. FDA puts it just a few lines above as [FDA] "will take compliance actions when appropriate, such as when the tests pose significant public health concerns."

And yes, you can still go on SNPedia and enter https://www.snpedia.com/index.php/Aspirin for example, they've just removed the pharmacogenetic section from the reports. They're still violating every other category of the FDA genetic tests charter, IFF research platforms/literature-retrieval systems are covered. It's a bit tricky for Promethease though, because of course they like to sell it as a complete & best consumer solution. I understand the argument that FDA's policies are unfair and stifle innovation, though in reality they hold Promethease's competition back lot more than them. (Well okay MyHeritage doesn't seem to care about that).

I have reservations about Dante Labs marketing and legal strategy, but that doesn't currently affect the product, nor is it my problem. However, I think the non-existent/fine-print disclaimers, especially on marketing materials, are much more common choice right now. Obviously they shouldn't be. On the other hand, on social media for example everybody(that's expressing an opinion for some reason) is pushing Promethease as a completely unerring, perfect analysis. There's active hate if someone says it's derived from publicly editable Wiki, contains errors, and shouldn't be used blindly (Although this does go for essentially every genetic, or biological test out there). Either practically nobody has read the disclaimers, or they choose not to believe them. I think Promethease reports do not contain any disclaimers themselves, and it was impossible to re-read them easily last I checked, so it's believable that the average customer doesn't even notice them.

But yeah, this too is getting into wider ethical discussions than the topic ;)

06-13-2019, 07:30 AM

06-14-2019, 04:56 AM
Well, that did not work out for me unfortunately. Dropbox seems to have an upper filesize limit of 50Gb. My BAM File exceeds this - its 98GB. Did you compress or split your file perhaps ? Or was it under 50GB ?

* edit * Well, figured it out. upload of 98Gb BAM file from PC to Dropbox in the app sucessfull. My bad - tried uploading it via internet - dumbass. Now i can finally drop the DL link to Promethease.