Method for analyzing loss of heterozygosity (LoH) following deterministic restriction site whole genome amplification (DRS-WGA)

Abstract

A method for analyzing loss of heterozygosity (LoH) in at least one sample containing genomic DNA is disclosed, the method comprising the following steps: a. providing at least one sample containing genomic DNA; b. performing deterministic restriction site whole genome amplification (DRS-WGA) of the genomic DNA; c. preparing a massively parallel sequencing library from the product of the DRS-WGA; d. performing low-pass whole genome sequencing on the massively parallel sequencing library with an average coverage depth of <1; e. aligning the reads obtained in step d on the reference genome of the at least one sample; f. extracting the allele content of multiple loci, wherein the multiple loci include polymorphic loci and / or heterozygous loci; g. assigning a LoH score to at least one genomic window of the reference genome for the at least one sample based on the number of loci having at least two different alleles among the multiple loci.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This patent application claims the priority of Italian Patent Application No. 102019000013335, filed on July 30, 2019, the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0003] The present invention relates to methods for analyzing loss of heterozygosity (LoH) in samples from low-pass whole genome sequencing data from deterministic restriction site whole genome amplification (DRS-WGA) with or without normal controls, thereby achieving single-cell resolution. The method can be applied to several single-cell applications, such as oncology, including analysis of single-cell heterogeneity in circulating tumor cells and tissue samples; or reproductive medicine, including preimplantation genetic screening (PGS). Background Art

[0004] Whole genome amplification (WGA) of single cell genomic DNA is often required to obtain more DNA to simplify and / or allow different types of genetic analysis including sequencing, SNP detection, etc. WGA based on deterministic restriction sites using LM-PCR (hereinafter referred to as DRS-WGA) is known from WO2000 / 017390.

[0005] Importantly, DRS-WGA has been shown to be a best-in-class WGA method in many aspects, especially with respect to lower allele dropout in single cells (Borgstrom et al., 2017; Normand et al., 2016; Babayan et al., 2016; Binder et al., 2014).

[0006] Hodgkinson CL et al., Nature Medicine 20, 897–903 (2014) used a commercial LM-PCR-based DRS-WGA kit (Ampli1 TM In this work, copy number analysis was performed by low-pass whole-genome sequencing of single-cell WGA material, followed by digestion and fragmentation with WGA adapters, followed by barcoded Illumina adapter ligation for sequencing.

[0007] WO2017 / 178655 and WO2019 / 016401A1 teach simplified methods for preparing massively parallel sequencing libraries from DRS-WGA (e.g. Ampli1) or MALBAC for low-pass whole genome sequencing and copy number analysis. The performance of the method of WO2017 / 178655 using the Ion Torrent platform has been described in detail with reference to copy number analysis in Ferrarini et al., PLoS ONE 13(3):e0193689 https: / / doi.org / 10.1371 / journal.pone.0193689.

[0008] Ampli1 TM WGA is compatible with array comparative genomic hybridization (aCGH). Indeed, several groups (Moehlendick B et al., 2013, PLoS ONE 8(6):e67031; Czyz ZT, et al., 2014, PLoS ONE 9(1):e85907) have shown that it is suitable for high-resolution copy number analysis. However, aCGH technology is expensive and labor-intensive, so that different methods may be needed, such as low-pass whole genome sequencing (LPWGS) for detecting somatic copy number alterations (CNAs).

[0009] DRS-WGA has been shown to better analyze copy number profiles from microdissected FFPE material than DOP-PCR when using array CGH, metaphase CGH, and other genetic analysis assays such as loss of heterozygosity using targeted primers and PCR for analysis of selected microsatellites (Stoecklein et al., Am J Pathol. 2002 Jul; 161(1):43-51; Arneson et al., ISRN Oncol. 2012; 2012:710692. doi: 10.5402 / 2012 / 710692. Epub 2012 Mar 14.).

[0010] US 7,424,368 B2 teaches a method for estimating the copy number of a genomic region in an experimental sample, the method comprising analyzing SNPs using a microarray. Microarray technology is less manipulable and flexible in terms of next generation sequencing, and does not provide absolute counts but only relative signals. In addition, there are setup costs associated with probe synthesis and microarray manufacturing, which is contrary to next generation sequencing (NGS).

[0011] Zahn H. et al., Nature Methods, Vol. 14, pp. 167-173 (2017) teach a method for preparing massively parallel single-cell libraries without pre-amplification and demonstrate simultaneous inference of CNAs and LoH for bulk-equivalents of the SA501X3F cell line. However, this method requires a relatively large number of single cells (48). In addition, in order to use TITAN for analysis, the position of heterozygous SNPs must be determined (Ha G. et al., 2014, Genome Research 24 (11)).

[0012] This approach has the following disadvantages.

[0013] 1. It is not compatible with the use of whole genome amplified libraries, but WGA is indeed required in many cases, such as when dealing with CTCs, because different aliquots of the WGA product may need to be re-analyzed to obtain additional information, such as information about SNVs in oncogenes or tumor suppressor genes, information at the single-cell level for each individual cell, for different purposes including for biomarker discovery or for evaluating other known biomarkers whose validity may not be inferred by low-pass WGS alone.

[0014] 2. In some applications, such as preimplantation genetic screening (PGS) or preimplantation genetic diagnosis (PGD), only single cells may be available, so the method of Zahn et al. is obviously not applicable.

[0015] 3. In some applications, there may be multiple cells available for analysis, but they may still not be enough to provide enough information to use the method of Zahn et al. For example, the number of CTCs collected from 7.5 ml of blood drawn from metastatic patients using the CELLSEARCH system was less than 10 in most cases (Allard WJ. et al., 2004, Clin Cancer Res., Oct 15; 10(20): 6897-904, see Table 2).

[0016] In oncology, genome-wide assessment of LoH has been shown to be important in several contexts, including assessment of the so-called BRCAness signature, which correlates with the efficacy of platinum therapy and poly(ADP-ribose) polymerase (PARP), an inhibitor in several cancer types (e.g., Watkins et al., Breast Cancer Research 2014, 16:211). In addition, LoH analysis of the BRCA1 and BRCA2 loci in tumors of individuals with germline mutations has been shown to be important for treatment efficacy.

[0017] In preimplantation genetic screening (PGS) or preimplantation genetic diagnosis (PGD), there is a need to assess uniparental disomy (UPD), which occurs when an individual receives two copies of a chromosome or part of a chromosome from one parent and no copy from the other parent. However, standard LPWGS workflows are unable to provide such information when using traditional bioinformatics pipelines and analysis methods.

[0018] It would be desirable to provide a method that allows inference of genome-wide LoH status (and / or gene-specific LoH status) to single-cell resolution that overcomes one or more of the following limitations inherent in prior art techniques:

[0019] - Requires high-coverage whole-genome sequencing, or equivalently, large quantities of single-cell low-pass sequencing to generate bulk equivalents with high coverage;

[0020] - Mandatory requirement for normal controls;

[0021] - Individual cells cannot be reliably reanalyzed for validation or to obtain additional genomic information of interest.

[0022] For CTC analysis, as well as for other single-cell analysis applications such as prenatal diagnosis of blastocysts and circulating fetal cells collected from maternal blood, efficient methods will be needed that combine the reproducibility and quality of DRS-WGA with the ability to analyze genome-wide LoH as well as copy number variations (CNVs) from the same low-pass sequencing data.

[0023] In addition, there is a need to determine genome-wide copy number profiles and LoH from minute amounts of cells, FFPE, or tissue biopsies.

[0024] Binder V. et al., "A new workflow for whole-genome sequencing of single human cells", Human mutation, Vol. 35, No. 10, pp. 1260-1270, 2014 discloses a workflow combining a WGA method based on efficient adapter-joint PCR with second-generation sequencing. This method allows comparison of single cells at base pair resolution. However, the method is based on genotyping SNPs (i.e., polymorphic genomic positions for which sufficient coverage is available) in order to call genotypes with a certain degree of confidence.

[0025] The above methods have significantly different objectives from the methods of the present invention. The objective of the present invention is not to genotype polymorphic positions as Binder et al. do, but to infer genome-wide LoH status (and / or gene-specific LoH status) to single-cell resolution.

[0026] The method of Binder et al. implies a number of reads that is two orders of magnitude larger. In contrast, according to the present invention, LoH can be called from a single sample, for example starting from 2 million reads, which corresponds to less than 1% of the reads used by Binder et al. Summary of the invention

[0027] Therefore, an object of the present invention is to provide a method for analyzing LoH which overcomes the disadvantages of the prior art methods.

[0028] In particular, the present invention aims to provide a method for analyzing LoH from a small number of cells to single-cell resolution after whole genome amplification, which method involves using fewer cells for analysis, fewer normal controls, and fewer sequencing reads per cell than generally reported in the art.

[0029] This object is achieved by a method as defined in claim 1 . BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A block diagram summarizing the steps of the method according to the invention is shown.

[0031] Figure 2 The effect of reduced genome representation on observed coverage is shown.

[0032] Figure 3 A diagram showing in-silico digestion of the human genome at different restriction sites is shown.

[0033] Figure 4 Computer analysis of the number of DNA fragments according to the variation in fragment length obtained by digestion of the hg19 human genome with Ampli1 DRS-WGA enzyme (MseI) is shown.

[0034] Figure 5 The relationship between the spacing and size of the fragments selected for sequencing is shown.

[0035] Figure 6 Schematic diagram showing an example of partitioning based on constant base pair genomic windows.

[0036] Figure 7 A schematic diagram showing an example of partitioning based on a constant number of loci per window.

[0037] Figure 8 is a graph showing the average number of heterozygous loci detected at different numbers of reads (1 to 9 million) in a genomic window of n=1000 SNPs. The vertical black line shows the standard deviation from the mean.

[0038] Fig. 9 A schematic diagram of the partitioning based on copy number bins is shown.

[0039] Fig.10 Shown is the case where the LoH score is defined as the number of heterozygous loci.

[0040] Fig.11 Shown is the case where the LoH score is defined as the proportion of heterozygous loci to the total number of loci in a genomic window with a constant number of loci.

[0041] Fig.12 Shown are ROC analyses for defining LoH score thresholds as the number of biallelic SNPs in a window of (A) n=500, (B) n=1000, (C) n=1500 SNPs covered by at least 1 read with 1,000,000 mapped reads.

[0042] Fig.13 Shown are the area under the receiver operating characteristic (ROC) curve (AUC) values ​​of the LoH score at different numbers of mapped reads (1,000,000-10,000,000 reads) and sample purities (10%-90%).

[0043] Fig.14 Shown are data related to fragment size selection obtained from sequencing libraries prepared using Illumina's Ampli1 LowPass.

[0044] Fig.15 An example of detecting LoH by using a constant base pair genomic window and applying Fisher's statistical test is shown.

[0045] Fig.16 An example of detecting LoH by using a genomic window with a constant number of loci is shown.

[0046] Fig.17 Shown is a ROC curve created based on a training set of 9 single cells for identifying the LoH score threshold, which is defined as the number of heterozygous loci in a window of n=1000 SNPs covered by at least 1 read with 1,000,000 mapped reads.

[0047] Fig.18A diagram showing a single tumor cell and LoH genomic region detected according to the method in Example 2 of the present invention.

[0048] Fig.19 An example of using copy number segmentation and Fisher's test to detect genomic regions with LoH status is shown.

[0049] definition

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by those of ordinary skill in the art to which the invention belongs. Although many methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, preferred methods and materials are described below. Unless otherwise indicated, the techniques described herein for the present invention are standard methods well known to those of ordinary skill in the art.

[0051] The expression "massively parallel next generation sequencing (NGS or MPS)" refers to a method for sequencing DNA that involves creating a library of spatially and / or temporally separated, clonally sequenced DNA molecules (or without prior clonal amplification). Examples include the Illumina platform (Illumina Inc), the Ion Torrent platform (Thermo Fisher Scientific Inc), the Pacific Biosciences platform, the MinIon (Oxford Nanopore Technologies Ltd).

[0052] The expression "low-pass whole genome sequencing" refers to whole genome sequencing performed with an average sequencing depth of less than 3 with reference to the entire reference genome.

[0053] The expression "average sequencing depth" herein means the total number of bases sequenced and mapped to the reference genome divided by the total reference genome size on a per sample basis. The total number of bases sequenced and mapped can be approximated as the number of mapped reads times the average read length.

[0054] The expression "reference genome" refers to a reference DNA sequence for a particular species.

[0055] The term "locus" (plural "loci") refers to a fixed position on a chromosome (relative to a reference genome).

[0056] The expression "polymorphic locus" refers to a locus having two or more alleles observed at a frequency greater than 1% in a population.

[0057] The expression "heterozygous locus" refers to a locus having two or more alleles observed in a particular sample.

[0058] The expression "genomic window" refers to an interval of a reference genome contained in a single chromosome, which has a fixed or variable length.

[0059] The expression "genomic region" refers to an interval comprising one or more adjacent genomic windows in the same chromosome.

[0060] The expression "covered genome" means the part of a reference genome that is covered by at least one read.

[0061] The term "read" refers to a piece of DNA that is sequenced ("read") by a sequencer.

[0062] The expression "copy number region" means a genomic region associated with the same copy number value.

[0063] The expression "segmented copy number regions" means genomic regions associated with the same copy number value as a result of bioinformatics analysis of CNAs.

[0064] The expression "tumor suppressor gene" refers to a gene that is associated with an increased probability of tumorigenesis due to loss of its function, for example, due to sequence variants (germline or somatic).

[0065] The expression "reduction rate" refers to the total number of bases included in the fragments within a specific base pair range obtained by in silico digestion of the reference genome with the restriction enzyme used in DRS-WGA divided by the total number of bases in the reference genome.

[0066] The expression "loss of heterozygosity" or "LoH" means that one of the alleles in a genomic region is missing.

[0067] The expression "LoH call" refers to the assignment of the presence of LoH (in a genomic region).

[0068] The expression "allele content" means the composition in terms of alleles detected at a locus.

[0069] Unless otherwise noted, for simplicity, in the description of the present invention, if only one allele is detected, the locus will be interchangeably referred to as homozygous or monoallelic, and in the case where at least two alleles are present, the locus will be referred to as heterozygous or biallelic, regardless of the actual genotype of the locus. DETAILED DESCRIPTION

[0070] refer to Figure 1, the method for analyzing loss of heterozygosity (LoH) in at least one sample containing genomic DNA according to the present invention comprises the following steps.

[0071] In step a, at least one sample comprising genomic DNA is provided.

[0072] In step b, defined restriction site whole genome amplification (DRS-WGA) of the genomic DNA is performed.

[0073] In step c, a massively parallel sequencing library is prepared from the product of the DRS-WGA.

[0074] In step d, the massively parallel sequencing library is subjected to low-pass whole genome sequencing at an average coverage depth of <1, preferably <0.05, and more preferably <0.01.

[0075] In step e, the reads obtained in step d are aligned to the reference genome of the at least one sample.

[0076] In step f, the allele contents at a plurality of loci are extracted. The plurality of loci include polymorphic loci and / or heterozygous loci.

[0077] In step g, a LoH score is assigned to at least one genomic window of the reference genome for the at least one sample according to the number of loci in the plurality of loci having at least two different alleles.

[0078] Preferably, a size selection step is performed before, during or after step c of preparing a massively parallel sequencing library, and the step of preparing a massively parallel sequencing library does not include a random fragmentation step.

[0079] The size selection step preferably retains fragments in the range of 100 to 800 base pairs.

[0080] In certain embodiments of the invention, size selection preferably retains fragments in the range of 300 to 450 base pairs.

[0081] In certain embodiments of the present invention, the peak of the fragments retained in the size selection step is preferably concentrated in the base pair range of 150 bp to 600 bp, and more preferably, the size selection step retains fragments in the range of 425 to 575 base pairs.

[0082] Preferably, at least one genomic window:

[0083] - have a constant base pair width, or

[0084] - has a constant number of said plurality of loci, or

[0085] - selected from the group consisting of a chromosome, a chromosome arm, and a segmented copy number region.

[0086] The plurality of loci preferably comprises polymorphic loci obtained from a reference genome for the at least one sample or a database (eg dbSNP) obtained by genotyping a set of reference samples.

[0087] Alternatively, the plurality of loci preferably comprises known heterozygous loci for a control sample.

[0088] When the genomic window has a constant base pair width or has a constant number of multiple loci, or the multiple loci include polymorphic loci for a reference genome for the sample, the LoH score preferably corresponds to the number of heterozygous loci in the at least one genomic window.

[0089] Preferably, the LoH score corresponds to the proportion of heterozygous loci to the total number of polymorphic loci in at least one genomic window.

[0090] The LoH score preferably corresponds to the p-value of a statistical test.

[0091] The statistical test preferably assesses the significance of over-representation of biallelic loci relative to sequencing and WGA error rates or the significance of under-representation of biallelic loci relative to control samples.

[0092] The control sample preferably comprises at least one genomic region of the major ploidy from at least one sample.

[0093] The control sample is preferably at least one normal sample, which is more preferably obtained from the same individual to be tested from which the at least one sample is obtained. In the case of oncology, the control sample is preferably a normal (non-tumor) sample.

[0094] In the case of circulating fetal cells, the control sample is preferably a maternal sample. Alternatively, when a paternal sample is available, it can be a paternal sample or a combination of a maternal sample and a paternal sample. The availability of maternal and / or paternal genotypes can be used to select a subset of loci that are known to be heterozygous in the parental control.

[0095] Preferably, if the LoH score exceeds (passes) a threshold value for the genomic window, the genomic window is referred to as being in LoH. In this case, the method more preferably comprises: a step of assigning a LoH state to each genomic window contained in at least one genomic region if the LoH score of the region exceeds the threshold value; or a step of assigning a LoH state to the region according to the LoH state of the genomic window contained in at least one genomic region.

[0096] More preferably, the at least one genomic region comprises a tumor suppressor gene, which is even more preferably selected from the group consisting of BRCA1, BRCA2, PALB2, TP53, CDKN2A, RB1, APC, PTEN, CDKN1B, DMP1, NF1, AML1, EGR1, TGFBR1, TGFBR2 and SMAD4.

[0097] The at least one sample preferably has a purity of at least 50%.More preferably, the at least one sample is a single cell.

[0098] Loci of univocal relationships in fragment length in DRS-WGA

[0099] In more detail, the method according to the invention exploits the fact that in DRS-WGA, such as Ampli TMWGA, each locus in the genome is represented in the WGA library only as fragments of a specific base pair length. This property can be designated as the "locus to fragment length univocal relationship" (L2FLUR). Considering a typical normal locus, such as the locus of a polymorphic SNP, the locus will be represented only by fragments of a given length, which is equal to the size of the corresponding fragment after restriction enzyme digestion (measured on either single strand), plus twice the universal WGA adapter length (i.e., the length of the LIB1 primer in the case of Ampli1 WGA). When the WGA is sequenced after library preparation according to the Ampli1 LowPass kit, a predictable additional length is introduced related to the known sequencing adapter and barcode lengths.

[0100] Non-ideal factors such as undigested restriction sites or sequence variants and other factors may affect and distort the representative frequency of a given fragment in the WGA product relative to theoretical expectations. These factors are usually modest, and as long as they are reproducible, their non-random nature can be partially offset by compensating for their effects. Therefore, unless otherwise stated, their effects will be ignored in this specification.

[0101] Reduced representation of the genome

[0102] In the method according to the invention, the L2FLURL property is exploited to produce a representative reduction of the genome, whereby, for a given number of reads, the low-pass sequencing data achieves an effectively higher coverage of the covered genome by effectively reducing the size of the covered genome relative to the original size of the sample reference genome. In other words, the size selection of the WGA fragments produces a deterministic subsampling of the reference genome. The term "deterministic" is essential because - increasing the number of reads - ultimately resampling the same genomic sites (see Figure 2 ).

[0103] Figure 2 The effect of reduced genome representation on observed coverage is shown. Figure 2 A shows the distribution of MseI fragment lengths according to three different methods: Ampli1 LowPass (A1LP_ss) for Ion Torrent with size selection to collect fragments of 300 to 450 bp, Ampli1 LowPass (A1LP) produced by the sequencing step, and the library obtained after random fragmentation and sequencing after Ampli1 WGA (A1_wFrg) (Binder V et al., 2014). These three different methods represent different levels of reduction in genomic representation, from the most stringent A1LP_ss to A1_wFrg characterized by lack of selection. Figure 2 B shows the Lorenz curves obtained using different methods, showing that the coverage uniformity gradually decreases with the size selection level. The lower uniformity of A1LP_ss can be explained by the saturation of the DNA template and repeated sequencing of the same fragment. Template saturation is determined by Figure 2 This is confirmed by the plots in C and 2D, which show the total amount of covered bases and the average coverage per base, respectively, at incremental intervals of mapped reads. These plots clearly show that the size selection step (A1LP_ss) reduces the amount of available DNA, with the effect of limited coverage targets, but with higher coverage.

[0104] It is worth noting that the method is flexible because different deterministic enzymes may be appropriate depending on the desired resolution and / or sequencing platform and sequencing protocol used. For example, different frequent cutters can be used. In the example of Ampli1 WGA, the TTAA motif is the restriction site. Other four-base cutters can be used to cut at different restriction sites, such as GTAC, CTAG ( Figure 3 ), thereby obtaining different fragment distributions. Figure 3In silico digestion of the human genome with different restriction sites (four or six base pairs) is shown. For a given range of fragment lengths (e.g., appropriate for a certain sequencer and size selection method), different restriction sites will produce different numbers of fragments.

[0105] When DRS-WGA is first purified after the primary PCR, a first size selection is performed, thereby removing shorter fragments of WGA together with free primers. Advantageously, the method uses a further selection step. This additional selection step can be achieved by selecting the size of certain fragments from the primary WGA and / or generating a massively parallel sequencing library by limiting the sequenceable fragments. For example, the Ampli1LowPass kit contains an inherent size selection step that is sufficient to have a positive impact on the method. In WO2017 / 178655, the gel is size-selected. In WO2019 / 016401, a continuous purification step using SPRI beads effectively produces a first size selection, whereby the length of the base pair is limited to a range that depends essentially on the concentration of the SPRI beads. In addition, the sequencer itself can also introduce size selection, because longer fragments will generate sequence data with increasingly lower efficiency (for example, due to the efficiency of emulsion PCR in Ion Torrent, or bridge PCR for cluster formation in the Illumina platform).

[0106] In DRS-WGA, there is also a deterministic relationship between the average size of sequencing libraries and the downsampling rate of the reference genome.

[0107] In silico analysis of TTAA digestion of the human reference genome hg19 ( Figure 4 ) produced a total of about 19M fragments including all chromosome sequences, which would translate into 38M fragments on a normal diploid human genome. For example, the computer selected fragments within the range of 175 to 225bp would have only 1,252,559, which covers a total of about 248M bases in 3.09B bases, i.e. 8.02% of the human reference genome. See Table 1 below, where the number of fragments, total base pairs, and reduction rate (%) are listed for different ranges selected by size. This downsampling can be specified as a reduction rate (RR).

[0108] Table 1

[0109] Spacing between segments

[0110]

[0111] In addition to the reduction rate, there is also a definite relationship between the average spacing between consecutive fragments in DRS-WGA, which depends on the part of the fragment length distribution selected for sequencing. In this regard, see Figure 5, where Figure A shows measurements of three different fragment sizes of 200, 500, and 800 with a band of ±100 bp. Due to the reduction in the number of selected fragments, there is a positive correlation between fragment length and spacing; Figure B shows that for each fragment size, three different bands (±50, ±100, ±150) are used to demonstrate that there is still a negative correlation between band size and spacing due to the reduction in the number of available fragments with a narrower size range.

[0112] In general, through computer analysis of the distribution of Ampli1 DRS-WGA fragments in the human reference genome hg19, it was found that:

[0113] ●The higher the average base pair length of the selected fragments, the smaller the number of fragments and the larger the spacing between them;

[0114] ●The narrower the selected fragment range, the fewer the fragments are and the greater the spacing between them.

[0115] Fragment size selection

[0116] Different size selection techniques can also be used to achieve the desired reduction ratio, depending on the number of sequencing reads selected per sample and / or the resolution. Figure 4 , it is obvious that, for a given average fragment length, selecting smaller or larger bands centered on that average fragment length, respectively, can result in a smaller or larger number of total fragments.

[0117] Instruments such as the Pipping prep (Sage Science) can be used to obtain tighter control over fragment length distribution and, using passband filter analogs, can also have higher Q factors, which are defined as

[0118] Q=Fcenter / DeltaF=[(Fmin+FMAX) / 2] / (FMAX-Fmin)

[0119] in

[0120] Fcenter = (Fmin + FMAX) / 2 is the average size of the fragment

[0121] DeltaF = FMAX - Fmin is the width of the fragment size range

[0122] Fmin is the fragment size below which fragments are expressed at a conventional relative level (eg 1 / 10 = 10%) or smaller relative to the normalized in-band peak fragment number in each bin.

[0123] FMAX is the fragment size above which fragments are expressed at conventional relative levels or smaller relative to the normalized in-band peak fragment number per bin.

[0124] For Illumina sequencing, paired-end sequencing is preferred as the sequencing mode increases because the genome covered increases, and thus the number of loci per million read pairs increases, thereby improving resolution. However, when the size selected for sequencing is below a certain size, paired-end sequencing does not increase coverage because the two paired reads completely overlap.

[0125] With Ion Torrent sequencing, higher read lengths will proportionally increase the genome covered, and therefore the number of loci per million reads, thereby improving resolution. In the Ampli1 LowPass IonTorrent Kit (Menarini Silicon Biosystems), barcoded pooled samples are size-selected on a gel or using other methods such as Pippin Prep. Selecting different Q factors and average fragment lengths can provide different resolutions on a per million read basis.

[0126] One advantage of pooling samples and selecting library sizes for subsequent sequencing is that all samples will have the same fragment length distribution, which in turn will maximize the overlap of the genome covered in different samples. This is relevant when using control-based methods (e.g., normal controls or maternal controls) to identify potential heterozygous loci in the sample to be tested (SUT).

[0127] On the other hand, when using the Ampli1 LowPass kit for Illumina, different LowPass libraries are first size-selected and then pooled, resulting in slightly different size selections in different samples, thus reducing the genome covered in different samples on a per million reads basis. Although not mandatory in standard protocols, size selection after library pooling can be used to increase overlap between samples, which may be beneficial for control-based analysis.

[0128] According to the present invention, the combination of DRS-WGA and LPWGS unexpectedly results in a reduced representation of the input sample. This reduced representation library of the reference genome in turn reduces the genome covered within the selected (or any sequenceable) base pair range by sequencing using NGS, and achieves effectively higher coverage of the genome covered on a per million read basis compared to alternative WGA methods using random priming or random shearing.

[0129] According to the invention, this effect can be utilized in different ways depending on the situation.

[0130] An example is the availability of one or more control samples - e.g. a "matched normal" and one or more samples under test (SUT) (e.g. a tumor sample). In this case, DRS-WGA increases the read overlap between the SUT and the control.

[0131] Another example is the case of no controls, such as in preimplantation genetic screening (PGS), where only a single sample corresponding to the SUT is available. In this case, DRS-WGA increases the number of loci covered by more than one read.

[0132] Preferably, the preparation of the library from DRS-WGA is one of the methods disclosed in WO2017 / 178655 and WO2019 / 016401, since the resulting reduction is higher than digesting the WGA adapters, fragmenting the DNA and subsequently creating a sequenceable library as performed in Binder V et al., 2014 or Hodgkinson CL et al., 2014. In fact, DNA shearing increases the number of possible different fragments of the original DRS-WGA that can be found in a given base pair range selected for sequencing, since once fragmented, longer fragments will fall back into the above range, whereas due to fragmentation, only a small fraction of the primary WGA fragments that were originally in the range will be eliminated from the range, since smaller fragments tend to be sheared less efficiently relative to longer fragments (see Figure 2 ).

[0133] LoH Analysis

[0134] Reference again Figure 1, the massively parallel sequencing library is preferably obtained using the Ampli1 LowPass kit (for IonTorrent or for Illumina). The sample is sequenced using a compatible sequencer. The sequencing reads obtained from the library are mapped to the reference human genome, and the alleles present in the known loci and / or polymorphic loci are extracted. Preferably, such loci are covered by at least 2 sequencing reads. It should be noted that the detection of a single allele does not necessarily mean a true homozygous genotype, but may be the result of low sequencing coverage. The multiple loci are preferably subdivided in the genome window according to different standards of genome partitioning. This partitioning is optional, because in some embodiments, people may only be interested in the analysis of one or several predetermined genome windows, such as a single chromosome or a single genome locus containing one or more genes of interest. The allele state of the locus detected in the genome window is used to obtain a measurement value. This measurement value, hereinafter referred to as the LoH score, can be obtained by a variety of methods according to the present invention, such as counting the number of heterozygous loci in the genome window, or calculating the proportion of heterozygous loci. In addition, statistical tests are preferably applied to determine the significance of the drop in heterozygous loci corresponding to LoH events, either by comparison with internal controls or by using external controls (from the same individual or from different individuals). Alternatively, statistical tests are preferably applied to determine the significance of the over-representation of heterozygous loci corresponding to genomic regions that are not in LoH relative to expectations based on sequencing and WGA error rates. Finally, LoH score thresholding is preferably applied to define genomic regions corresponding to LoH events, based on a fixed threshold calculated from a training data set with known LoH events. The individual steps of the method are described in detail below.

[0135] Genome partitioning

[0136] refer to Figure 1 , the optional step of partitioning can be performed in three alternative ways:

[0137] i) Constant base pair genomic window

[0138] ii) Constant number of locus windows

[0139] iii) Copy number segment

[0140] exist Figure 6In the alternative i shown, the genomic window has a constant width. Each genomic window comprises multiple loci, and its quantity depends on the genomic position. When a sample is compared with a group of control normal samples, this method may be advantageous, because the reference genome is partitioned in the same way in all samples, thereby allowing the LoH score of each genomic window to be directly compared across multiple samples. Due to the number and ratio of the heterozygous loci detected in the genomic window of the defined width will increase under a higher read depth, to allow the sample to be compared with the control (or multiple) control samples, the number of mapped reads in each sample is preferably standardized to a fixed number of reads. This standardization is performed by randomly sampling reads, mapped to the reference genome, until the required number of times is reached. The number of reads standardized can be, for example, 1,000,000 or 2,000,000 reads, preferably 3,000,000, 4,000,000, 5,000,000, 6,000,000, 7,000,000, 8,000,000 or 9,000,000 reads.

[0141] Figure 6 A schematic diagram of an example of constant base pair based genomic window partitioning is shown. Paired control (top) and test (bottom) samples are shown. The solid line represents (a portion of) the genome. Diamond markers delimit genomic windows of constant width, and known polymorphic loci are represented by dots (heterozygous loci: white filled dots; homozygous loci: gray filled dots). The number of loci detected by each genomic window varies across the genome, but on average, for a given window, is expected to be similar between two different samples whose total read mappings have been normalized to a defined read count. Genomic windows in LoH in the test sample are expected to show a decrease in heterozygous loci compared to the same window in the normal control sample. Due to biases in SNP density along the genome, the same window cannot be directly compared to genomic windows located at different genomic locations on the same (or other) samples.

[0142] exist Figure 7 In the alternative ii) shown, the genomic window has a constant number of loci. This method allows standardization of the LoH score for different SNP densities across the entire genome. When using an uncontrolled method, the method may be advantageous because it allows, for example, the same threshold to be applied to all genomic windows, regardless of their position in the genome and their potential SNP density. When a test sample is compared with a control sample, the method may be disadvantageous because different samples may generate different genomic windows, depending on the loci distribution detected by sampling and low-pass sequencing.

[0143] Figure 7A schematic diagram of an example of a constant number of partitions based on each window locus is shown. Paired control (top) and test (bottom) samples are shown. The solid line represents (a portion of) the genome. The diamond marker defines a genomic window containing a constant number of loci. Known polymorphic loci are represented by dots (heterozygous loci: white filled dots; homozygous loci: gray filled dots). Due to the low sequencing coverage, all loci in the genomic region will not be detected. Therefore, the genomic window ends can vary between different samples based on the sampling of loci performed by sequencing reads, and therefore, the genomic windows detected in the test sample cannot be directly compared to the corresponding genomic windows in other (control) samples. Relative to the genomic windows of the same sample that are not in LoH, the genomic windows in LoH in the expected test sample show a decline in heterozygous loci.

[0144] The number and proportion of heterozygous loci detected in a genomic window with a constant number of loci increases with increasing read depth (see Figure 8 ). Preferably, to allow thresholding of the LoH score to a pre-calculated value, the number of mapped reads in each sample is normalized to a fixed number of reads. This normalization is performed by randomly sampling reads and mapping to the reference genome until the desired number of reads is reached. The normalized number of reads can be, for example, 1 million or 2 million reads, preferably 3 million, 4 million, 5 million, 6 million, 7 million, 8 million or 9 million reads.

[0145] In such Fig. 9In the alternative iii) shown, the genomic window is a segmented genomic region between two copy number breakpoints, which is contained in a chromosome arm and can be defined by normalizing the raw copy number counts in the genomic window according to GC content (Boeva, V. et al., 2011, Bioinformatics, 27 (2), 268-269) and by applying a segmentation algorithm such as an algorithm based on LASSO (Harchaoui, Z. et al., 2008, Adv. Neural Inform. Process. Syst., 20, 617-624), circular binary segmentation (CBS) (Seshan VE. et al., 2019, DNAcopy: DNAcopy number dataanalysis. R package version 1.58.0) or similar algorithms for normalizing read counts. This method is based on the assumption that genomic regions showing copy number level changes relative to the main "normal" ploidy of the sample may be affected by a single genomic copy number aberration event and are therefore expected to have a uniform LoH state. Compared to alternatives (i) and (ii), the genomic window defined by this method is typically much larger (up to 2-3 orders of magnitude) and will contain a large number of known heterozygous and / or polymorphic loci, so higher statistical power can be achieved. In addition, by combining 2 different biological dimensions (copy number, LoH score), more accurate results and lower false positive rates can be obtained using this method. However, this method may be disadvantageous in the case where smaller LoH events that are still undetectable by this method are located in larger copy number events. Since it is not uncommon for chromosome arms to experience LoH events after replication, it is preferred that chromosome arms be used as segmentation units in chromosomes without changing the copy number. This prevents the shorter chromosome arms from being mistakenly called as LoH when only the longer arms are affected (false positive), or in double cases, when only the shorter chromosome arms are affected, the shorter chromosome arms are mistakenly called as no-LoH (false negative).

[0146] More specifically, Fig. 9 An exemplary representation of the copy number profile of a chromosome arm (genomic major ploidy = 2) affected by two copy number change events is provided: copy number loss segment, copy number = 1; copy number gain, copy number = 3. The genomic window is defined as the region between 2 consecutive copy number breakpoints.

[0147] Segmentation can also be used to use copy number information to exclude false positives originating from high-level amplification. In fact, high-level amplification is most likely to originate from a single allele, thus introducing a bias in the allele representation in this region, so that the minor allele (even if present) will be underrepresented and may lead to false positive LoH calls.

[0148] Table 2 below shows the main features and advantages and disadvantages of each alternative partitioning step according to the invention.

[0149] Table 2

[0150]

[0151] LoH score

[0152] The step g of assigning a LoH score to at least one genomic window of the reference genome for the at least one sample according to the number of loci having at least two different alleles in the plurality of loci also relates to an alternative preferred embodiment.

[0153] In a preferred embodiment, the LoH score corresponds to the number of heterozygous loci in the at least one genomic window. Genomic windows in LoH are expected to show a scarcity of heterozygous loci compared to regions or samples not in LoH (see Fig.10 ).

[0154] In another preferred embodiment, for each genomic window, the LoH score is defined as the ratio of heterozygous loci detected in that genomic window relative to the total number of polymorphic loci in the same genomic window ( Fig.11 ). Similar to the above method, in the presence of LoH events, the LoH score is expected to continue to decrease. This method may be advantageous when the window does not contain a homogenous number of loci detected, such as when a constant base pair genomic window is used or when copy number segments are used to partition the genome.

[0155] LoH score - statistical test

[0156] Preferably, for each genomic window, the LoH score is defined by the results of a statistical test of the observed frequencies of biallelic loci.

[0157] In a preferred embodiment, the significance of the low representation of heterozygous loci relative to internal / external controls can be evaluated by performing statistical tests. Specifically, the following two classifications are considered to establish a contingency table for each genomic window: 1) sample type (test, control); 2) locus type (heterozygous, homozygous). Statistical tests are then applied, such as Fisher's exact test or alignment test (e.g., chi-square test, G test, Barnard's exact test, Fisher-Freeman-Halton test) for analyzing the contingency table. Preferably, the statistical test should be performed on one side so as to limit the detection to the case where the heterozygous loci are underrepresented due to LoH. In fact, when there is a gain in a given genomic segment, that is, when the copy number increases, the number of reads using low-pass WGS will increase. This may result in more heterozygous loci in the absence of LoH, and may be marked as significant by a two-sided statistical test but for reasons contrary to the purpose of the analysis.

[0158] In an alternative preferred embodiment, the over-representation of heterozygous loci can be tested for expected significance relative to sequencing and WGA error rates. This approach may be advantageous when testing "gain of heterozygosity" (hereinafter referred to as GoH) in haploid single cells (e.g., gametes). This may occur, for example, due to errors in unbalanced segregation that lead to chromosomal gain during meiosis.

[0159] Given the large number of tests performed for each experiment (approximately 200, 400, 600 times for a 1 million read sample with a fixed window of 500, 1000 and 1500 SNPs), multiple testing correction can be applied (see, e.g., Benjamini Y et al., 1995, Journal of the Royal Statistical Society. Series B (Methodology) Vol. 57, No. 1: pp. 289-300). The LoH score is then defined as the p-value derived from the statistical test.

[0160] Control samples

[0161] The control may be "internal" and may be defined, for example, by considering genomic regions whose ploidy is equal to the most likely primary (average) genome ploidy. This approach assumes that most genomic regions that do not show copy number changes are not in LoH.

[0162] Alternatively, a control may be "external" and may be generated, for example, by using one or more normal samples from the same individual being tested or from a different individual.

[0163] The use of internal controls can be advantageous for diploid or polyploid samples (e.g. tumor samples) since it is not dependent on the number of reads (no normalization on the number of mapped reads is required) and in cases where the samples are damaged (e.g. FFPE samples). In fact, damaged samples may show a higher occurrence of dropouts, where one of the two alleles of a locus is lost due to DNA damage, compared to undamaged samples, and therefore a lower than expected number of heterozygous sites for genomic regions that are not in LoH. This may hinder the comparison of tests with external control samples with different degrees of damage. By using an internal control, this bias is eliminated, since the control and test genomic windows will have the same level of dropouts.

[0164] LoH Threshold and LoH Call

[0165] Optionally, the LoH scores obtained from the previous step can be thresholded to define genomic regions in LoH. In most cases, the number and proportion of heterozygous loci with a constant number of loci detected in a genomic window will increase with increasing read depth. In order to allow the LoH score to be thresholded to a pre-calculated value, the number of mapped reads in each sample is preferably normalized to a fixed number of reads. This normalization is performed by randomly sampling reads and mapping to the reference genome until the required number of times is reached (preferably contained in the range of 1,000,000 mapped reads to 10,000,000 mapped reads). The above considerations do not apply when the LoH score is calculated by performing statistical tests on "internal" controls.

[0166] Preferably, in cases where the LoH score is calculated as the number of heterozygous loci, the data is first downsampled to 1,000,000 mapped reads. Loci covered by at least 1 read are partitioned using windows of a fixed number of detected loci (e.g., n=500; n=1000; n=1500). Some preferred thresholds are 3, 6, 9 heterozygous SNPs in 500, 1000, and 1500 loci, respectively. Fig.12 ). If the LoH score is below the chosen threshold, LoH is called in a given genomic window.

[0167] In more detail, Fig.12 ROC analysis for defining LoH score thresholds is shown, which are defined as the number of bi-allelic SNPs in a window of (A) n=500, (B) n=1000, (C) n=1500 SNPs covered by at least 1 read with 1,000,000 mapped reads. LoH detected in tumor cells by high-throughput whole genome sequencing and B allele frequency analysis was used as a reference.

[0168] In case the LoH score is calculated as a p-value, due to the statistical test applied, some preferred thresholds may be, for example, 5*10-2 or 1*10-2. If the LoH score is below the selected threshold, LoH is called in the genomic window.

[0169] Once the LoH score reaches a threshold, the LoH status can be assigned to the genomic region according to different criteria described below.

[0170] 1) LoH region calling by merging windows. In this preferred embodiment, if the LoH score of each genomic window contained in the genomic region passes the thresholding step, the LoH state is assigned to the genomic region.

[0171] 2) LoH region calling based on LoH state in genomic windows. In the preferred embodiment, LoH state is assigned to a genomic region if a given percentage / portion of genomic windows contained in the genomic region passes the thresholding step. For example, if more than 66%, 75%, 80%, 85%, 90%, 95% of the windows in the genomic region pass the thresholding step, LoH state is assigned to the genomic region.

[0172] 3) LoH calling in genomic regions containing tumor suppressor genes. In this preferred embodiment, at least one genomic region contains a tumor suppressor gene.

[0173] Preferably, the gene is selected from BRCA1, BRCA2, PALB2, TP53, CDKN2A, RB1, APC, PTEN, CDKN1B, DMP1, NF1, AML1, EGR1, TGFBR1, TGFBR2 and SMAD4.

[0174] Sample purity

[0175] LoH can be identified in DNA from a mixture of different cell types (e.g., tumor cells and normal cells). Sample purity is defined as the percentage of the sample in the mixture that is of the type of interest (e.g., tumor cells).

[0176] For example, when clonal #TC tumor cells (i.e., genomically identical and therefore having the same LoH and CNA patterns) are mixed with #NC normal cells from the same individual, the purity of the resulting sample will be #TC / (#TC+#NC) and will be uniform across the genome.

[0177] In general, purity here refers to a concept related to the LoH status in a given region of interest that contains one or more genomic regions. The region of interest may be as large as the entire reference genome (as in the example above) or as small as 100kbp.

[0178] For example, in the presence of a pool of tumor cells representing different clones derived from the same last common ancestor tumor cell, when the LoH status of the genomic region is ubiquitous in all clones derived from the last common ancestor, the purity of different genomic regions may vary from a minimum of 1 / the number of cells in the pool (when the LoH region is represented in only one cell) to a maximum of 100%.

[0179] The sample analyzed for LoH preferably has a purity of at least 50%, more preferably at least 70%, as can be understood from 13. Fig.13 The receiver operating characteristic (ROC) area under the curve (AUC) values ​​for the LoH score at different mapped reads (1000000-10000000 reads) and sample purity (10%-90%) are shown. The LoH score is defined as the number of heterozygous SNPs in a window of n=150 SNPs covered by at least 2 reads. Samples of different purity were obtained by mixing in silico reads obtained from the analysis of tumor cells and normal cells at a ratio equivalent to the target purity (tumor:normal). The LoHs detected in tumor cells by high-throughput whole genome sequencing were used as a reference.

[0180] Effect of size selection on LoH detection

[0181] As mentioned above, size selection is preferably performed during or after step c of preparing the massively parallel sequencing library. The size of the fragments can be selected according to different criteria. The sequencing method can be selected by different criteria, which also depends on the fragment size. In general, the higher the number of loci (polymorphic or heterozygous) that contribute to the LoH analysis, the better the resolution (per million reads).

[0182] Fig.14 Shown are data obtained by in silico selection of a subset of sequencing fragments from data obtained from real single cell samples (sequencing libraries prepared using Ampli1 for Illumina's LowPass) with added Fcenter. Fig.14 A shows the effect of size selection (bandwidth 100) on DRS-WGA fragment coverage relative to the average fragment length (with 250,000 reads); Fig.14 B shows the effect of size selection (bandwidth 100) on the resolution in terms of base pairs (windows of 150 SNPs covered by at least 2 reads) with 250,000 reads; Fig.14C shows the effect of size selection bandwidth on DRS-WGA fragment coverage at a fixed average fragment length (500 bp) (with 250,000 reads); Fig.14 D shows the effect of size selection bandwidth on resolution (bp) at a fixed average fragment length (500 bp) (with 250,000 reads); Fig.14 E shows the effect of the number of reads on DRS-WGA fragment coverage at a fixed average fragment length (500 bp). The ratio of fragments covered by at least 2 reads to the total number of fragments covered increases in proportion to the number of mapped reads (dashed line); Fig.14 F shows the effect of the number of reads on the resolution (bp) at a fixed average fragment length (500 bp).

[0183] These data indicate that the total number of DRS-WGA fragments decreases while the number of fragments covered by more than one read useful for calling SNPs increases, reaching a plateau at 500 bp ( Fig.14 A). The resolution is thereby increased, as shown by the decrease in the length of the genomic window with a fixed number of SNPs (n=150; Fig.14 B). When different bandwidths are applied to a given number of mapped reads and Fcenter, the coverage and resolution of fragments increase as the bandwidth decreases ( Fig.14 C and 14D). Resolution also increases with the number of mapped reads ( Fig.14 E and 14F).

[0184] Example

[0185] Table 3 below summarizes the characteristics of the methods used in the three examples disclosed below.

[0186] Table 3

[0187]

[0188] Example 1

[0189] In Example 1, Ampli1 LowPass for Illumina DNA libraries of 1 circulating tumor cell (CTC; test) and 1 white blood cell (WBC; control) obtained from a male patient affected by multiple myeloma was considered. Sequencing reads were mapped to the hg19 reference human genome and downsampled at 1, 2, 3, 4, 5, 6, 7, 8, 9 million reads. Alleles at dbSNP polymorphic loci (dbSNP150 common variants with minor allele frequency ≥ 5%) were extracted from both libraries. The loci were segmented with a fixed 10000000 bp genomic window. A one-sided Fisher's exact test was used to assess the significance of the association between the two classifications (Table 4), with the null hypothesis that heterozygous and homozygous loci in WBC (control) and CTC (test) were likely to be the same.

[0190] Table 4

[0191]

[0192] The test results for each downsampling level are shown in Fig.15 Starting with 2 million reads, the method showed high sensitivity in detecting known LoH events on chromosomes 11 and 13.

[0193] In detail, Fig.15 The top figure shows a copy number map of CTCs from patients with multiple myeloma. The x-axis is chromosome; the y-axis is copy number. Each point represents a genomic window of fixed size. Copy number segments are represented by solid lines. The reference (Ref) track is shown below the copy number map, which represents the known LoH region detected by high-pass whole genome sequencing of the same CTC, shown in pure black. Below is the track marked 1M to 9M: a heat map of the logarithmic p-value (base = 10) of the Fisher's exact test results for different numbers of reads (1 to 9 million). More significant values ​​are represented by darker gray shades.

[0194] Example 2

[0195] In Example 2, the same single CTC data used in Example 1 was used as input and the data was downsampled at 1 million reads. In this case, the loci were partitioned into a fixed number (n=1000) of loci covered by at least 1 read in the window. For the identification of LoH regions, the LoH score was calculated as the number of heterozygous positions in each window.

[0196] Fig.16Detection of LoH by using a genomic window with a constant number of loci is shown. In particular, the top figure shows the copy number graph of the same CTC as in Example 1. The x-axis is the chromosome; the y-axis is the copy number. Each point represents a genomic window of fixed size. The copy number segments are represented by solid lines. Below the figure is a heat map showing the heterozygous counts for each genomic window. Windows with lower LoH scores (fewer heterozygous loci) are more likely to be in a LoH state, represented by darker gray shading. Chromosome 11, the large arm of chromosome 13, and chromosome X (single copy in male individuals) show lower LoH scores.

[0197] To determine the LoH score threshold for calling genomic windows in the LoH state, a training set of 9 single cells with known LoH regions was analyzed using the same method as the test samples (1,000,000 mapped reads and n = 1,000 SNP windows). ROC analysis was then performed, and the maximum LoH score threshold = 6 was determined to be the best compromise between sensitivity and specificity ( Fig.17 , where the x-axis represents 1-specificity (lower values ​​indicate more specific detection) and the y-axis represents sensitivity. LoH detected in tumor cells by high-throughput whole-genome sequencing was used as a reference).

[0198] The approach successfully identified LoH events on chromosomes 11 and 13. LoH status was also assigned to chromosome X, as expected in male individuals whose genomes contain a single copy of the X chromosome ( Fig.18 - Regions with LoH scores below a fixed threshold (<=6) and larger than 10,000,000 bp are shown in black).

[0199] Example 3

[0200] In Example 3, Ampli1 LowPass for Illumina libraries of 2 single Hodgkin Reed / Sternberg (HRS) cells obtained from FFPE tissues of classical Hodgkin lymphoma samples of male patients was analyzed. Two HRS cells share the same copy number spectrum. Sequencing reads were mapped to the hg19 reference human genome, and alleles present at the dbSNP polymorphic locus (dbSNP150 common variants with minor allele frequency ≥ 5%) were extracted from the two libraries. The locus was partitioned using the copy number segments obtained by using Control-FREEC software to achieve GC-based standardization and copy number signal segmentation [Boeva, V et al., Bioinformatics, 27 (2), 268-269. http: / / doi.org / 10.1093 / bioinformatics / btq635). An internal control defined by the union of all regions where the copy number is equal to the cell ploidy (copy number = 2) was used. For each segment defined by copy number analysis and contained in a chromosome arm, a one-sided Fisher's exact test was performed to reject the null hypothesis that the observed biallelic and monoallelic loci were most likely identical in the segment and in the internal control ( Fig.19 - Top panel: a representative HRS cell copy number profile. Bottom panel: heatmap of -log10 of p-values ​​obtained as output of Fisher's test. Only genomic regions with p-value ≤ 0.01 are shown. More significant values ​​are represented by darker grey shading). As expected, all regions with copy number = 1 were correctly detected as LoH genomic regions. The long arm of chromosome X was detected in the LoH state despite copy number = 2. This was expected since the sample was from a male individual and thus the genome contained a single X chromosome. In addition, chromosome 9q was called in LoH, which would have been missed using only copy number information (copy number = 2).

[0201] advantage

[0202] The method according to the present invention is suitable for analyzing data obtained from low-pass sequencing of genomic DNA of a test sample to detect LoH events. In contrast to other methods that infer LoH as the run of continuous homozygous loci and require extraction of true genotypes at a certain number of loci, the method of the present invention is based on the following principle: by analyzing a genomic window containing a sufficient number of loci sequenced at low coverage, and by extracting the alleles observed at said loci (not necessarily representative of the sample genotype), LoH events can be detected as a reduction of biallelic loci compared to what is observed by analyzing a normal diploid sample.

[0203] In contrast to other methods that infer LoH from alternative allele frequencies (B allele frequency or BAF), which require high coverage of the genome, e.g. 30x (Boeva ​​et al., Bioinformatics, Vol. 28 no. 3 (2012), pp. 423-42), the method according to the present invention works with low-pass whole genome sequencing data (<1x or lower, down to e.g. 0.05x or even 0.01x), with corresponding cost savings.

[0204] The method for analyzing LoH from a sample according to the present invention allows the use of very few samples to infer LoH regions of the entire genome from low-pass whole genome sequencing data to single-cell resolution, because in cases where there may be very few samples, only a few (as low as one) CTCs are available, with an additional optional possibility of running the analysis without normal controls and with a relatively small number of reads.

[0205] Furthermore, certain embodiments of the method are able to improve the resolution of LoH calls by introducing certain processing steps during library preparation without increasing sequencing costs.

[0206] The method according to the invention surprisingly improves the prior art with performance that was previously considered unachievable by a person skilled in the art. In particular, the method allows:

[0207] - Identification of LoH on single cells by low-pass whole-genome sequencing with average coverage as low as 0.01-0.04 (250,000 to 1,000,000 single-end 150 bp reads for the human genome);

[0208] - Obtain the above points without control samples;

[0209] - The above points are obtained, and further the possibility to obtain additional genetic material for studying other characteristics of the single cells, as well as the possibility to reliably reanalyze single cells for validation purposes, due to the use of inherent WGA in the process.

[0210] Furthermore, the method according to the invention allows the determination of genome-wide copy number profiles and LoH even from minute amounts of cells, FFPE or tissue biopsies.

Claims

1. A method for analyzing loss of heterozygosity (LoH) in at least one sample comprising genomic DNA, wherein: The method is used for non-disease diagnosis purposes, and the method comprises the following steps: a. providing at least one sample comprising genomic DNA; b. performing deterministic restriction site whole genome amplification (DRS-WGA) of the genomic DNA; c. preparing a massively parallel sequencing library from the product of the DRS-WGA; d. performing low-pass whole genome sequencing on the massively parallel sequencing library with an average coverage depth of <1; e. aligning the reads obtained in step d on the reference genome of the at least one sample; f. extracting the allele content of multiple loci, wherein the multiple loci include polymorphic loci and / or heterozygous loci; g. assigning a LoH score to at least one genomic window of the reference genome for the at least one sample according to the number of loci having at least two different alleles in the plurality of loci, A genomic window is a region of the reference genome contained in a single chromosome, which has a fixed or variable length. The allelic status of the loci detected in a genomic window is used to obtain a measurement, called the LoH score, which is obtained by counting the number of heterozygous loci in a genomic window, or calculating the proportion of heterozygous loci. Wherein, if the LoH score is below a threshold value for a genomic window, the genomic window is referred to as being in LoH, in which case the method comprises: a step of assigning a LoH state to each genomic window contained in at least one genomic region if the LoH score of the region is below the threshold value; or a step of assigning a LoH state to the region according to the LoH state of the genomic window contained in at least one genomic region, and Wherein, the LoH score threshold is defined as the number of biallelic SNPs in a window of (A) n=500, (B) n=1000, or (C) n=1500 SNPs covered by at least 1 read with 1,000,000 mapped reads.

2. The method according to claim 1, wherein: A size selection step is performed before, during or after step c of preparing a massively parallel sequencing library, and the step of preparing a massively parallel sequencing library does not include a random fragmentation step.

3. The method according to claim 2, wherein: The size selection step retains fragments between 100 and 800 base pairs.

4. The method according to claim 3, wherein: The size selection step retains fragments between 300 and 450 base pairs.

5. The method according to claim 3, wherein: The peak of the fragments retained in the size selection step is concentrated in the base pairs of 150 bp to 600 bp.

6. The method according to claim 5, wherein: The size selection step retains fragments between 425 and 575 base pairs.

7. The method according to claim 1, wherein: The at least one genomic window has a constant base pair width.

8. The method according to claim 1, wherein: The at least one genomic window has a constant number of the plurality of loci.

9. The method according to claim 1, wherein: The at least one genomic window is selected from the group consisting of a chromosome, a chromosome arm, and a segmented copy number region.

10. The method according to claim 1, wherein: The plurality of loci comprises polymorphic loci for a reference genome of the at least one sample.

11. The method according to claim 7, wherein: The LoH score corresponds to the number of heterozygous loci in the at least one genomic window.

12. The method according to claim 10, wherein: The LoH score corresponds to the proportion of heterozygous loci to the total number of polymorphic loci in the at least one genomic window.

13. The method according to claim 10, wherein: The LoH score corresponds to the p-value of the statistical test.

14. The method according to claim 13, wherein: The statistical test assesses the significance of over-representation of biallelic loci relative to sequencing and WGA error rates.

15. The method according to claim 13, wherein: The statistical test assesses the significance of underrepresentation of biallelic loci relative to control samples.

16. The method according to claim 15, wherein: The control sample comprises at least one genomic region of the major ploidy from the at least one sample.

17. The method according to claim 15, wherein: The control sample is at least one normal sample.

18. The method according to claim 17, wherein: The at least one normal sample is obtained from the same individual to be tested from which the at least one sample is obtained.

19. The method according to claim 15, wherein: For the at least one sample, the control sample is a maternal sample or a paternal sample.

20. The method according to claim 1, wherein: The at least one sample has a purity of at least 50%.

21. The method according to claim 20, wherein: The at least one sample is a single cell.

22. The method according to claim 1, wherein: In step d, the massively parallel sequencing library is subjected to low-pass whole genome sequencing at an average coverage depth of <0.

05.

23. The method according to claim 1, wherein: In step d, the massively parallel sequencing library is subjected to low-pass whole genome sequencing at an average coverage depth of <0.01.

Images