# US 2025/0157573 A1 · Genome wide assembly-based structural variant calling

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0157573 A1
- Document ID: US20250157573A1
- Status: Published application
- Kind: A1
- Application number: 18/887,480
- Filed: 2024-09-17
- Published: 2025-05-15
- Earliest claimed: 2023-11-09
- Assignees: Illumina, Inc.
- Applicants: Illumina, Inc.
- Inventors: Camilla Colombo; Mitchell A. Bekritsky; Ritu Kundu; Stefan Dorokhov
- Web page: https://patentgrep.com/patent/US20250157573A1

## CPC

- G16B 20/20: Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
- G16B 20/10: Ploidy or copy number detection
- G16B 30/10: Sequence alignment; Homology search
- G16B 30/20: Sequence assembly
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY

## Abstract

This disclosure describes systems, non-transitory computer readable media, and methods for implementation of SV calling pipelines that can generate high-confidence diploid assemblies and SV call sets based on trio read data. For instance, a system is provided for implementing an SV calling pipeline via a bioinformatics subsystem executed on one or more processors that can receive a trio dataset of diploid read data of sequenced child (CH) and parental (M, P) genomes, and generate, based on consensus regions of the trio read data, a validated diploid assembly of the CH genome. Validation can be determined based on composition vectors of corresponding child and transmitting parent sequences at a given interval for each consensus region. The bioinformatics subsystem can then detect structural variants (SV) within the CH diploid assembly and output a VCF file of detected SVs as an SV truthset.

## Claims

### Claim 1

A system comprising:
at least one processor; and
a non-transitory computer readable medium comprising instructions that, when executed by the at least one processor, cause the system to:
receive a trio dataset of diploid read data of sequenced child and parental genomes;
generate, based on the read data, aligned consensus regions in haplotype-specific (phased) child and parental sets;
validate the child consensus regions at intervals (I), wherein validation for each consensus region is determined based on a distance between composition vectors of corresponding child and transmitting parent sequences at each interval;
output, in an alignment format, a datafile of validated child consensus regions as a diploid assembly;
detect structural variants (SV) within the diploid assembly; and
output, in a variant call format, a datafile of detected SVs as an SV call set.

### Claim 2 (depends on 1)

The system of claim 1, wherein composition vector is determined based on a summing of k-mer frequency vectors at a given interval.

### Claim 3 (depends on 2)

The system of claim 2, wherein (I) is between ˜105 and 1010 bp.

### Claim 4 (depends on 2)

The system of claim 2, wherein
the value k is 1≤k≤I and the frequency vector (v) is determined for each I−k+1 number (n) of k-strings to yield v1−n k-mer frequency vectors per interval (I).

### Claim 5 (depends on 4)

The system of claim 4, wherein k is between ˜3 and 102 bp.

### Claim 6 (depends on 4)

The system of claim 4, wherein a child consensus region at a given interval can be determined valid if the composition vector of each child sequence of respective haplotypes is within a threshold distance measure of the composition vector of one of the two sequences of respective haplotypes of a transmitting parent.

### Claim 7 (depends on 6)

The system of claim 6, wherein the distance measure is expressed as a similarity value between 0 and 1, given by the formula:
similarity
=
cos
⁡
(
θ
)
=
A
·
B

A

⁢

B

=
∑
i
=
1
n
⁢
A
i
⁢
B
i
∑
i
=
1
n
⁢
A
i
2
⁢
∑
i
=
1
n
⁢
B
i
2
,
wherein
A is the composition vector at a sequence one,
B is the composition vector at a sequence two,
A·B is a dot product of k-mer frequency vectors Av1−n at a sequence one and corresponding k-mer frequency vectors Bv1−n at a sequence two, and
∥A∥ ∥B∥ is the Euclidean norm; and
wherein
sequence one is a haplotype-specific sequence of the child consensus region; and
sequence two is one of the two haplotype-specific sequences of the transmitting parent.

### Claim 8 (depends on 8)

The system of claim 8, wherein the threshold is ˜0.93 to 0.97.

### Claim 9 (depends on 6)

The system of claim 6, wherein validation is determined in two or more passes,
wherein the instructions, when executed by the at least one processor, cause the system to:
validate, in a first pass, each child consensus region at a first interval (I); and
validate, in a second pass, each region at interval (I) determined to be invalid in the first pass at a second subinterval (Is).

### Claim 10 (depends on 1)

The system of claim 1, wherein the SV call set comprises one or more SV classes, wherein each SV class is one or a combination of an insertion, deletion, inversion, duplication, or translocation of >50 bp.

### Claim 11 (depends on 10)

The system of claim 10, wherein large segment deletions and duplications of >1000 are classified according to copy number variation (CNV).

### Claim 12 (depends on 1)

The method of claim 1, wherein the alignment format datafile comprises a set of genomic coordinates of the diploid assembly.

### Claim 13 (depends on 10)

The method of claim 10, wherein the variant call format datafile comprises, for each detected SV, a representation of SV class and location in the diploid assembly.

### Claim 14 (depends on 1)

The system of claim 1, wherein the instructions that, when executed by the at least one processor, cause the system to:
output, in respective alignment format datafiles, corresponding tandem repeat (TR) and non-tandem repeat (NTR) datasets of validated child consensus regions;
detect SVs within the NTR dataset;
detect CNV within the TR dataset;
output, in a variant call format, a merged SV call set of detected SVs within the NTR dataset and detected CNV within the TR dataset.

### Claim 15 (depends on 1)

The system of claim 1, wherein the instructions that, when executed by the at least one processor, cause the system to:
validate, in a first pass, each child consensus region at interval (I);
generate, from regions determined not to be valid in pass 1, split sets of TR regions and NTR regions; and
validate, in second pass, the split set of TR regions and, separately, the split set of NTR regions at a subinterval (Is).

### Claim 16 (depends on 1)

The system of claim 1, wherein the instructions, when executed by the at least one processor, cause the system to:
perform a phenotype study based, at least in part, on one or both of the alignment format and variant call format datafiles.

### Claim 17 (depends on 1)

The system of claim 1, wherein the instructions, when executed by the at least one processor, cause the system to:
perform a benchmark analysis of a variant call set based, at least in part, on one or both of the alignment format and variant call format datafiles.

### Claim 18 (depends on 1)

The system of claim 1, wherein the instructions, when executed by the at least one processor, cause the system to:
construct or revise a reference genome based, at least in part, on one or both of the alignment format and variant call format datafiles.

### Claim 19

A method of generating an SV truth set from a trio dataset, the method comprising:
receiving a trio dataset of diploid read data of sequenced child and parental genomes;
generating, based on the read data, aligned consensus regions in haplotype-specific (phased) child and parental sets;
validating the child consensus regions, wherein validation for each consensus region is determined based on composition vectors of corresponding child and transmitting parent sequences at a given interval (I);
outputting, in an alignment format, a datafile of validated child consensus regions as a diploid assembly;
detecting structural variants (SV) within the diploid assembly; and
outputting, in a variant call format, a datafile of detected SVs as an SV call set.

### Claim 20 (depends on 19)

The method of claim 19, wherein composition vector is determined based on a summing of k-mer frequency vectors at a given interval.

### Claim 21 (depends on 20)

The method of claim 20, wherein (I) is between ˜105 and 1010 bp.

### Claim 22 (depends on 20)

The method of claim 20, wherein
the value k cannot violate 1≤k≤I and the frequency vector (v) is determined for each I−k+1 number (n) of k-strings to yield v1-n k-mer frequency vectors per interval (I).

### Claim 23 (depends on 22)

The method of claim 22, wherein k is between ˜3 and 102 bp.

### Claim 24 (depends on 22)

The method of claim 22, wherein a child consensus region at a given interval can be determined valid if the composition vector of each child sequence of respective haplotypes is within a threshold distance measure of the composition vector of one of the two sequences of respective haplotypes of a transmitting parent.

### Claim 25 (depends on 24)

The method of claim 24, wherein the distance measure is expressed as a similarity value between 0 and 1, given by the formula:
similarity
=
cos
⁡
(
θ
)
=
A
·
B

A

⁢

B

=
∑
i
=
1
n
⁢
A
i
⁢
B
i
∑
i
=
1
n
⁢
A
i
2
⁢
∑
i
=
1
n
⁢
B
i
2
,
wherein
A is the composition vector at a sequence one,
B is the composition vector at a sequence two,
A·B is a dot product of k-mer frequency vectors Av1-n at a sequence one and corresponding k-mer frequency vectors Bv1-n at a sequence two, and
∥A∥ ∥B∥ is the Euclidean norm; and
wherein
sequence one is a haplotype-specific sequence of the child consensus region; and
sequence two is one of the two haplotype-specific sequences of the transmitting parent.

### Claim 26 (depends on 25)

The method of claim 25, wherein the threshold is ˜0.93 to 0.97.

### Claim 27 (depends on 24)

The method of claim 24, wherein validation further comprises
wherein the instructions, when executed by the at least one processor, cause the system to:
validate, in a first pass, each child consensus region at s first interval (I); and
validate, in a second pass, each region at interval (I) determined to be invalid in the first pass at a second subinterval (Is).

### Claim 28 (depends on 19)

The method of claim 19, wherein the SV truth set comprises one or more SV classes, wherein each SV class is one or a combination of an insertion, deletion, inversion, duplication, or translocation of >50 bp.

### Claim 29 (depends on 19)

The method of claim 19, wherein large segment deletions and duplications of >1000 are classified according to copy number variation (CNV).

### Claim 31 (depends on 28)

The method of claim 28, wherein the variant call format datafile comprises, for each detected SV, a representation of SV class and location in the diploid assembly.

### Claim 32 (depends on 19)

The method of claim 19 further comprising:
outputting, in respective alignment formats, corresponding tandem repeat (TR) and non-tandem repeat (NTR) datasets of validated child consensus regions;
detecting SVs within the NTR dataset;
detecting CNV within the TR dataset; and
outputting, in a variant call format, a merged SV truth set of detected SVs within the NTR dataset and detected CNV within the TR dataset.

### Claim 33 (depends on 29)

The method of claim 29, wherein the alignment format datafile comprises a set of genomic coordinates of the diploid assembly.

### Claim 33 (depends on 19)

The method of claim 19, wherein validating further comprises:
validating, in a first pass, each child consensus region at interval (I);
generating, from regions determined not to be valid in pass 1, split sets of TR regions and NTR regions; and
validating, in second pass, the split set of TR regions and, separately, the split set of NTR regions at a subinterval (Is).

### Claim 34 (depends on 19)

The method of claim 19 further comprising:
performing a benchmark analysis of a variant call set based, at least in part, on the SV truth set.

### Claim 35 (depends on 19)

The method of claim 19 further comprising:
constructing or revising a reference SV call set, at least in part, on the SV truth set.
