PDA

View Full Version : blood sample option for 10x Chromium



FGC Corp
03-29-2019, 02:39 AM
Using blood samples, FGC can achieve 21,000,000 Callable Loci.
Sample result: 21,066,571

We have identified an option for storing and shipping blood samples at room temperature. Blood samples can be used for 10x Chromium long read tests.

Táltos
03-29-2019, 05:01 AM
Using blood samples, FGC can achieve 21,000,000 Callable Loci.
Sample result: 21,066,571

We have identified an option for storing and shipping blood samples at room temperature. Blood samples can be used for 10x Chromium long read tests.

How will blood samples be obtained? Venipuncture in a lab? Fingerstick at home?

FGC Corp
03-29-2019, 05:12 AM
How will blood samples be obtained? Venipuncture in a lab? Fingerstick at home?

Thus far, customers are going to labs to get blood draws and then returning the blood samples to our lab. Fingerstick may be doable at some point.

WSO
03-30-2019, 03:09 PM
Blood is better than saliva?

JamesKane
03-30-2019, 03:18 PM
Blood is better than saliva?

You can see the impact of a blood sample versus saliva samples in the details of the Y-DNA Warehouse BAM coverage comparison. https://ydna-warehouse.org/coverage.html.

9T7D6 is from a blood sample. The other Chromium examples were saliva.

If you’re going to invest in cutting edge testing, you will want do as much as you can to ensure quality fragments are available for sequencing.

Snarple
04-08-2019, 12:18 AM
To be fair 9T7D6 is also ~60x coverage https://my.pgp-hms.org/profile/huA2692E the quality isn't due to blood alone. WGS compressed BAM is 112GB.

JamesKane
04-08-2019, 10:39 AM
Yes, extra average depth of coverage helps but there is a rate of diminishing returns. A traditional 50x WGS sample is only 400,000 more callable bases than a 30x WGS test. FGC would have to use their internal numbers with other blood samples to share out if that holds on the long read tests.

dtvmcdonald
04-08-2019, 05:30 PM
FullGenomes is claiming 19,877,569 callable loci for my saliva LR sample, which was 50X or a bit more.

HOWEVER I have my own design calling program which has proven extremely reliable when
comparing numerous related files (R1a-YP330) by various sources, including BigY500, BigY700,
various FullGenomes products, and a couple of other WGS products. This program only works on
haploid genomes.

It gives me 21,608,004 callable loci compared to Build38 reference. To that add about 100,000
loci callable in the DYZ19 region using a de-novo Pacbio reference.

My calling program has serious disagreements with FullGenome's one about which loci
are "callable" but, compared over many people, "unreliable". These are in the high tens of thousands, not millions.
In general these are in regions where the Long Reads are not "helping" as much as their alignment
program thinks they are, for purposes of getting a good mapping quality.

Several of the loci my programs calls but theirs does not have proven to be "golden" or "platinum"
for genealogical purposes.

JamesKane
04-08-2019, 05:45 PM
I believe FGC is using an unmodified version of GATK3’s callable loci tool for this metric. Build versions of the tool will cause small differences. If you look at the module’s documentation, https://software.broadinstitute.org/gatk/documentation/tooldocs/3.8-0/org_broadinstitute_gatk_tools_walkers_coverage_Cal lableLoci.php, you can peruse what the numbers represent.

You may disagree with Broad Institute here, and that is fine as long as you’re consistent in your analysis.

MacUalraig
04-08-2019, 05:58 PM
I've played around with the CallableLoci function at home. It is worth pointing out though that 4, the magic number we discussed before, is just a default figure.

When I've gathered up the rest of my pending BAM files I'll write up all the stats from both GATK and YFull.

FGC Corp
04-10-2019, 02:57 AM
I believe FGC is using an unmodified version of GATK3’s callable loci tool for this metric. Build versions of the tool will cause small differences. If you look at the module’s documentation, https://software.broadinstitute.org/gatk/documentation/tooldocs/3.8-0/org_broadinstitute_gatk_tools_walkers_coverage_Cal lableLoci.php, you can peruse what the numbers represent.

You may disagree with Broad Institute here, and that is fine as long as you’re consistent in your analysis.

GATK is among the best of the tools available.

dtvmcdonald
04-10-2019, 08:15 PM
I think that the reason my caller is generating more callable loci is that before a read is used at all
for any other test, it is examined to see how many differences it has with the reference. If the
read has more than a certain number of differences, not counting indels, it is discarded. This is
essentially a prefilter for read quality in certain regions, especially the region above 50 million,
and also certain repeat regions (but not DYZ19) and parts of palindromic regions. This causes
MORE calls because it cuts out loci where, otherwise, one would get heterogeneous calls
in what should be a haploid genome. The usual number I use is 6, though 4 seems fine too, and 8 is
clearly too large.

Examination of BAM files
shows that this is a reasonable test, PLUS .... many of the SNPs called by using this have proven
100% reliable using phylogenetic tests, and none have failed phylogenetic tests. Some smaller number
have proven actually useful (i.e. not Yet Another Synonym.)

FGC Corp
04-11-2019, 03:02 AM
I think that the reason my caller is generating more callable loci is that before a read is used at all
for any other test, it is examined to see how many differences it has with the reference. If the
read has more than a certain number of differences, not counting indels, it is discarded. This is
essentially a prefilter for read quality in certain regions, especially the region above 50 million,
and also certain repeat regions (but not DYZ19) and parts of palindromic regions. This causes
MORE calls because it cuts out loci where, otherwise, one would get heterogeneous calls
in what should be a haploid genome. The usual number I use is 6, though 4 seems fine too, and 8 is
clearly too large.

Examination of BAM files
shows that this is a reasonable test, PLUS .... many of the SNPs called by using this have proven
100% reliable using phylogenetic tests, and none have failed phylogenetic tests. Some smaller number
have proven actually useful (i.e. not Yet Another Synonym.)

So, what was the payoff for your project? I do understand that the payoff is limited by the lack of other 10x Chromium samples in your project.

dtvmcdonald
04-11-2019, 03:47 PM
Payoff 1) Finding out what the real nature of the big delete(s) in the DYZ19 region really are.

Payoff 2) Autosomal phasing. I've got 89% of my genome 100% phased into maternal and paternal parts.
The remaining 11% awaits more cousins' tests to provide data to assign maternal or paternal
sides to the LR phasing.

Payoff 3) two useful deletes (one single base, one 4 bases) were found there and verified in BigY500 BAMs.
For some reason I had never found these either in FTDNA data or my own calling of BigY BAMs. It is not
clear why I had not found them, or perhaps had fouind them but not noticed their usefulness, as they show
up clearly in the raw results from my caller on BigY data. They do depend on the "feature" of
my caller described in my recent post. Apparently
the spreadsheet I had generated to look through results visually did not visually highlight them properly.
I found them in investigating in detail all of your "new" indels.

But remember ... I paid the money for Payoff 1. I'm a scientist, and was just curious. In hindsight
I could have found a hint by trying to fit my plain Illumina data to the Build 38 "small extra fragment", but
would never have thought that that fragment could just possibly be right were I was looking. I might add that
where it is is now known, with respect to the bulk of DYZ19, and thus roughly where in Builld 38 it is.
(That is, it is near the real beginning of DYZ19). It was well worth the money.

pauleug
04-19-2019, 03:19 AM
How does it work in reality? Are there labs who have agreement with you? Who pays for the venipuncture?

FGC Corp
04-19-2019, 03:20 AM
How does it work in reality? Are there labs who have agreement with you? Who pays for the venipuncture?

The customer would need to set-up an appointment with a lab. We can provide the appropriate blood tube for this DNA test.