Gene deletion called as homozygous

[mate-ldw]

Dear developer,

We sequenced an FMR1 repeat expansion using a targeted PureTarget library. The TRGT analysis reported 28 CGG repeats, homozygous, and based on the assigned karyotype (XX), two alleles were detected.

However, subsequent cytogenetic analyses (karyotyping and FISH) revealed a terminal deletion on one X chromosome that includes the FMR1 gene.

We attempted additional analyses using hiPhase for phasing and Sawfish for structural variant detection, but due to the targeted nature of the PureTarget approach, these analyses did not provide any indication of the deletion.

Do you have any suggestions on how this limitation could be addressed? In particular, is there a recommended strategy for detecting large deletions affecting the target locus when using a targeted enrichment approach?

Best regards,
Max

[tmokveld]

Hello,

I cannot be 100% sure without seeing your data, but with targeted enrichment, if one X copy has a terminal deletion that deletes FMR1 and the breakpoint also lies outside the captured region, you will likely not have any breakpoint-spanning reads. Which would explain why an SV caller like sawfish will not necessarily detect the event when used with targeted data (note that sawfish can also do CNV detection, but this is out of scope for targeted data). Now TRGT will genotype the allele that is present, if only one allele is observed (or both alleles are identical at the repeat), the call can appear "homozygous".

One possible solution is doing copy number inference from sequencing depth, i.e., you can compare normalized on-target coverage at/around FMR1 to the other panel targets and, if possible, to a set of samples run with the same panel. A single copy in an XX sample should show a clear coverage drop. Additionally, you could check on the haplotype that is present if you have informative flanking SNVs in the FMR1 capture region, as the absence of heterozygosity could provide more supporting evidence.

Happy to discuss this in more detail or look at the data if you're able to share it.

[mate-ldw]

Dear Tom, Thank you very much for your response. That’s what I suspected as well — a deletion large enough to span the entire FMR1 gene would likely be missed by Sawfish. Are you suggesting an approach similar to what you’re implementing in the PTCP pipeline for SMN (https://github.com/PacificBiosciences/ptcp/tree/main/docker/ptcp/scripts/smn-homology)? I’ve attached the VCF files from TRGT and DeepVariant. Please let me know if you need any additional files or information. Best regards, Max Labor Dr. Wisplinghoff Dr. rer. nat. Maximilian Thelen Abteilung Molekulargenetik Horbeller Str. 18-20 50858 Köln Tel.: +49 221 940 505 607 Fax.: +49 221 940 505 412 ***@***.******@***.***> www.wisplinghoff.de<http://www.wisplinghoff.de/> ==================================================== Aus Rechts- und Sicherheitsgründen ist die in dieser E-Mail gegebene Information nicht rechtsverbindlich. Eine rechtsverbindliche Bestätigung reichen wir Ihnen gerne auf Anforderung in schriftlicher Form nach. Beachten Sie bitte, dass jede Form der unautorisierten Nutzung, Veröffentlichung, Vervielfältigung oder Weitergabe des Inhalts dieser E-Mail nicht gestattet ist. Diese Nachricht ist ausschließlich für den bezeichneten Adressaten oder dessen Vertreter bestimmt. Sollten Sie nicht der vorgesehene Adressat dieser E-Mail oder dessen Vertreter sein, so bitten wir Sie, sich mit dem Absender der E-Mail in Verbindung zu setzen. ---------------------------- For legal and security reasons the information provided in this e-mail is not legally binding. Upon request we would be pleased to provide you with a legally binding confirmation in written form. Any form of unauthorised use, publication, reproduction, copying or disclosure of the content of this e-mail is not permitted. This message is exclusively for the person addressed or their representative. If you are not the intended recipient of this message and its contents, please notify the sender immediately. ==================================================== Von: Tom Mokveld ***@***.***> Gesendet: Dienstag, 17. Februar 2026 10:28 An: PacificBiosciences/trgt ***@***.***> Cc: Thelen, Maximilian ***@***.***>; Author ***@***.***> Betreff: Re: [PacificBiosciences/trgt] Gene deletion called as homozygous (Issue #95) WARNUNG: Diese Mail wurde von extern zugestellt. Bitte nicht auf Links klicken oder Anhänge öffnen, wenn Sie den Absender nicht kennen! [https://avatars.githubusercontent.com/u/123967438?s=20&v=4]tmokveld left a comment (PacificBiosciences/trgt#95)<#95 (comment)> Hello, I cannot be 100% sure without seeing your data, but with targeted enrichment, if one X copy has a terminal deletion that deletes FMR1 and the breakpoint also lies outside the captured region, you will likely not have any breakpoint-spanning reads. Which would explain why an SV caller like sawfish will not necessarily detect the event when used with targeted data (note that sawfish can also do CNV detection, but this is out of scope for targeted data). Now TRGT will genotype the allele that is present, if only one allele is observed (or both alleles are identical at the repeat), the call can appear "homozygous". One possible solution is doing copy number inference from sequencing depth, i.e., you can compare normalized on-target coverage at/around FMR1 to the other panel targets and, if possible, to a set of samples run with the same panel. A single copy in an XX sample should show a clear coverage drop. Additionally, you could check on the haplotype that is present if you have informative flanking SNVs in the FMR1 capture region, as the absence of heterozygosity could provide more supporting evidence. Happy to discuss this in more detail or look at the data if you're able to share it. — Reply to this email directly, view it on GitHub<#95 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BLUX3XLQ5DUR53G34DEKN2T4MLNIFAVCNFSM6AAAAACVIT62RSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTSMJTGMZDEMBZG4>. You are receiving this because you authored the thread.Message ID: ***@***.******@***.***>>

[tmokveld]

If it is easier to share, you can contact me directly on: tmokveld@pacificbiosciences.com

Indeed it would resemble how it is done in for SMN in PTCP whenever there is full homology in the haplotypes.

[dnil]

This is bordering on a commercial, but since it's free and might help, here goes. If you have the possibility, consider running https://github.com/dnil/strdrop as well on the TRGT output VCF. It would not correct the genotype, but should flag the locus as potentially LowDepth and give some statistics to help guide further handling. We are still a bit early on the curve there so feedback would be appreciated. 😊