Question from Anton:

1.Can we get cut-off Z value for each specific chromosome?

Slide 16: T8 – Z score is close to 8 (cut off value of autosome trisomy is obviously different that of T21,
T13 and T18.

Just to be sure if we understood correctly: Z value up to 3-4 is considered grey area. Z value 4.5 is kind of
grey area also because it is near the limit value? Then we have to check other parameters to report positive
or we need to re-library immediately?

One example: I guess next to T5 a Z value is written?

Answer:

For chromosome 13/18/21, Z score from 1.96 to 4 is considered as grey area which will be suggested to re-
library at the 1st testing while for the other autosome, the cut off value is 8.

Z score such as 4.5 for T13/18/21 is close to the threshold, but normally the Z score will be a lot higher
than 4, and therefore it is better to re-library for verification. But please consider the other parameters (CNV
chart to see if there is fluctuation/ if it is GC outlier) when making decisions.

2. Z value more than 4 – please explain.

Bellow sample has Z value of more than 4.5 but your bioinformatics team suggested to relibrary, even
though HALOS marked is qualified.

Answer:

As mentioned in Question 1, for T13/18/21, HALOS will present T* (e.g. T21) when Z-socre is higher than
4, but it is suggested to re-library when Z-score reaching near the threshold.
 3. Sample 18B0287004D has fetal fraction below 3.5% - 2.86% actually with GC content 42.08% but
HALOS suggested to relibrary – why not re-sample immediatelly?

Answer: Either the sample quality or experiment condition could affect the fetal fraction.

Usually sample with low fetal fraction without GC outlier/GC content QC fail is suggested to resample
according to our experience. But with GC content failure, re-library will be preferred as sample quality
might not be the cause and direct resample will cause unease on the patients.

4. If we understood correctly, if maternal duplication/deletion is larger than 5M, you suggest to report it?

Following sample with maternal deletion: usually you recommend not to report it, but with sample bellow
we got discrepant results – to report it than NOT TO report it, and it’s not larger than 5M. Please advise.

This sample is maternal CNV, which does not exclude the possibility of fetal
Negative
del(13q21.1,1.97M)-M abnormalities. Feedback is recommended.

Answer:

CNV size over 5M without maternal effect is suggest to report

CNV size over 5M with maternal effect: the tested CNV result is suspected to be influenced by maternal
background. Before elimination of the maternal CNV by BGI, the suggestion will be “Negative for CNV”
because it is highly possible that the test result is affected by maternal background which means the baby’s
CNV is not able to be evaluated.
5. For slide 19 (Figure 9 in HALOS SOP). Highy repeated sequence in N region caused false positive. Will
it always cause false results around/near the N region or it can cause shifts somewhere far away on the
graph/chromosome arms? I can see that the “CNV” is inside grey/white stripes also.

Answer: Yes, based on our experience, CNV result showing reads lie in the N region will always causing
false positive result.

6. Also can you explain why on the next graph we do not think that the results are influenced by N region
repetitive sequence? Is it maybe that most of the deletion on the longer chromosome arm extends far away
from the N region beyond grey/white stripes? If I understood correctly: in yellow box we have false deletion
influenced by N region, in green box we have positive reported deletion that extends further?

Positive: del( 9q12-q31.1, 39.43 M)

7. We noticed in HALOS that some graphs for qualified low risk samples are not red, they are blue. Is this
related to lower reads for the whole chromosome? Do we have to pay attention to this?

Answer: if t-score for all the spots in CNV chart is below 2, then the line will become blue.

8. Regarding level A and level B Qc points – what is the acceptable level for low mapping rate – is it 60%,
65%?

Answer: The required mapping rate is 75 to 100 shown in the graph. As you know the level B QC points is
supplementary to level A, the result can be given for several samples in the patch with slightly low level B
indexes, but if the whole patch of sample are all with low mapping rate, then it is not acceptable and this is
supposed to cause the patch QC failure or lane QC failure.

9. Case with GC outer + fetal fraction below 8% -> HALOS didn’t offer us a relibrary but your manual
review did – this would have to be incorporated in the HALOS:
Your explanation was as follows: Re-library – because of GC outer

Answer: In the SOP, you can find 'GC outlier with < 8% fetal fraction and negative result. Suggest to re-
library for verification'. And for your request on the improvemet on HALOS， we will feedback to our
team for considering the changes in HALOS setting.

10. However on the second occasion we had two cases with GC outlier and fetal fraction below 8% and
you suggested a negative report:

Your explanation was the following: “The fraction of the two samples fit the required concentration
(≥3.5%) of HALOS, and GC outer doesn’t always lead to QC failed. There are some other data set
in the halos which is used together to judge the QC state. Like these two samples, you can send our
results export from halos for bioinformatic help.”

So my question: Which are those other data sets in the halos that are used for QC? Please explain last two
cases in detail so we can understand and make decisions on our own for the future.

Answer: Please check the sentence in the SOP: “Twice tests with GC outlier with < 8% fetal fraction and
negative result. Suggest to release negative report.” AND the reference data set in HALOS contains
confidential information and we are not at liberty to provide the actual information.

Question from Eva:

1. Aneupoidy fraction display for all chromosomes. Currently Nifty is the only test in the market that will report
rare autosomal trisomy/monosomy. Since the test is performed early in pregnancy, we can expect these cases, so we
would kindly ask for additional information. The doctors that have been asking us for such information would benefit,
also the patients in deciding whether to perform invasive procedures and which to perform. We would also be much
more competitive on the market in the area of special cases and could form a good relationship with the doctors so
that they could provide us feedback for special cases.
Answer：This is related to the characteristics of NIPT product and scope of BGI service, you can discuss this with
Albert.
2. Please incorporate in HALOS system a hierarchy for reporting advice. Example: if the sample has low fetal
fraction and 42.5% GC content, it will ask for re-library instead of re-sample. It would be useful for the system to
report re-sample as the „strongest“ action and then other re-library actions.

Answer: Please kindy check question 3 above.

3. Please enable pdf export/zipped folder for 1 patient/all patients with all graphs inside. This would be very
useful for tracking the CNV-s of individual chromosomes, as well as total sample quality. Also, it would be one-click
operation and faster to send and review.
Answer: After the bioininformatic training, we have reflected this request to the Halos research team and they have
already tried to realize this in the next Halos version.

4. Please include your references in the HALOS bioinformatics SOP. We saw them on your slides. It would be
nice to have them listed at the end of the SOP also. Could you also include studies relating chr 7 and chr 9 being often
false positive?
Answer: As soon as i get it from them, i send it to you.