discrepancy between pre-computed scores and spliceAI scores about spliceai HOT 4 CLOSED

Unfortunately, the precomputed files are available only for -D 50 setting.
…
On Sun, Mar 28, 2021 at 9:32 PM fjmuzengyiheng @.***> wrote: There are two reasons for this discrepancy: - In the paper, for each variant, we looked for splicing changes up to 50 positions on either side of the variant. In version 1.2.1 of the github software, we looked for 500 positions on either side as we were told by users that it catches more splice-altering variants. - In the data we provided (spliceai_scores.raw.snv.hg19.vcf.gz), we also filter out scores which correspond to increase in the strength of a canonical site and decrease in the strength of a non-canonical site, as these are not likely to cause rare genetic diseases. However, these differences have caused some confusion among some users. To avoid this, in Version 1.3, we have made the following changes: - We have provided an extra optional parameter D, which the users can set. If you set it to 50 (default), you will be able to reproduce the precomputed scores. - We have also added an extra optional parameter M, which if you set to 0, you get scores without any masking of scores, and if you set to 1, scores which correspond to increase in the strength of a canonical site and decrease in the strength of a non-canonical site will be filtered out. I would recommend you to download v1.3 of the software and the precomputed files here: https://basespace.illumina.com/analyses/194103939/files?projectId=66029966 You will no longer see any discrepancies. I apologize for the confusion. Sorry to be a bother. I am using VEP (--plugins spliceai) to annotate variants. However, it seems that precomputed files (spliceai_scores.raw.snv.hg19.vcf.gz) here were using -D 50(default). ( https://basespace.illumina.com/analyses/194103939/files?projectId=66029966 ) Are there precomputed files using -D much bigger (like -D 1000 or even -D 9999)? Anyway, thanks again for providing this powerful software. — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#26 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AE44RH7ZXBXWJU2XR3ABWNLTF5HHPANCNFSM4JCMDT7A .

hi, it there precomputed files for -D (more than 50) setting now?~ It will be good if we have this to avoid missing intronic variants in WGS data. Thank you again~!

from spliceai.

There are two reasons for this discrepancy:

• In the paper, for each variant, we looked for splicing changes up to 50 positions on either side of the variant. In version 1.2.1 of the github software, we looked for 500 positions on either side as we were told by users that it catches more splice-altering variants.
• In the data we provided (spliceai_scores.raw.snv.hg19.vcf.gz), we also filter out scores which correspond to increase in the strength of a canonical site and decrease in the strength of a non-canonical site, as these are not likely to cause rare genetic diseases.

However, these differences have caused some confusion among some users. To avoid this, in Version 1.3, we have made the following changes:

• We have provided an extra optional parameter D, which the users can set. If you set it to 50 (default), you will be able to reproduce the precomputed scores.
• We have also added an extra optional parameter M, which if you set to 0, you get scores without any masking of scores, and if you set to 1, scores which correspond to increase in the strength of a canonical site and decrease in the strength of a non-canonical site will be filtered out.

I would recommend you to download v1.3 of the software and the precomputed files here: https://basespace.illumina.com/analyses/194103939/files?projectId=66029966 You will no longer see any discrepancies. I apologize for the confusion.

from spliceai.

There are two reasons for this discrepancy:

• In the paper, for each variant, we looked for splicing changes up to 50 positions on either side of the variant. In version 1.2.1 of the github software, we looked for 500 positions on either side as we were told by users that it catches more splice-altering variants.
• In the data we provided (spliceai_scores.raw.snv.hg19.vcf.gz), we also filter out scores which correspond to increase in the strength of a canonical site and decrease in the strength of a non-canonical site, as these are not likely to cause rare genetic diseases.

However, these differences have caused some confusion among some users. To avoid this, in Version 1.3, we have made the following changes:

• We have provided an extra optional parameter D, which the users can set. If you set it to 50 (default), you will be able to reproduce the precomputed scores.
• We have also added an extra optional parameter M, which if you set to 0, you get scores without any masking of scores, and if you set to 1, scores which correspond to increase in the strength of a canonical site and decrease in the strength of a non-canonical site will be filtered out.

I would recommend you to download v1.3 of the software and the precomputed files here: https://basespace.illumina.com/analyses/194103939/files?projectId=66029966 You will no longer see any discrepancies. I apologize for the confusion.

Sorry to be a bother.
I am using VEP (--plugins spliceai) to annotate variants.
However, it seems that precomputed files (spliceai_scores.raw.snv.hg19.vcf.gz) here were using -D 50(default).
(https://basespace.illumina.com/analyses/194103939/files?projectId=66029966)
Are there precomputed files using -D much bigger (like -D 1000 or even -D 9999)?

Anyway, thanks again for providing this powerful software.

from spliceai.

from spliceai.