WO2023116717A1

WO2023116717A1 - Method for monitoring donar dna fraction

Info

Publication number: WO2023116717A1
Application number: PCT/CN2022/140435
Authority: WO
Inventors: Jianxing HE; Liping Liu; Liyan Huang
Original assignee: The First Affiliated Hospital Of Guangzhou Medical University
Priority date: 2021-12-22
Filing date: 2022-12-20
Publication date: 2023-06-29

Abstract

Provided is a method for detecting donor cfDNA fraction in a sample of a recipient. Also provided is a method for detecting over-represented SNP(s)of donor cfDNA in an identified region.

Description

METHOD FOR MONITORING DONAR DNA FRACTION

Introduction

Organ transplantation is now a preferred practice for patient with end-stage organ failure. However, the clinical outcomes remain poor with median survival rate. The immunological reaction of rejection is a major cause of functional failure in transplant patients. Accurate monitoring of graft is essential for long-term survival of the transplant recipient. The current "gold standard" for detecting or confirming graft rejection following organ transplantation requires biopsy samples (for example, transbronchial biopsy for lung transplant and endomyocardial biopsy for heart transplant) in order to detect immune cell (e.g., T-cells, macrophages, etc. ) infiltration into the graft and other pathological changes. These invasive techniques suffer from high cost and myriad complications.

The discovery of the positive correlation between the fraction of donor-derived cell-free DNA (cfDNA) in the recipient’s plasma (herein denoted as donor DNA fraction) and the risk of organ transplant rejection has empowered the development of non-invasive methods for the prediction and prevention of organ transplant failure. However, previous studies mainly focused on a global estimation of donor DNA fraction conducted months post-transplant, and there is a critical need for non-invasive detection methods that can be used in early detection of allograft rejection and prognosis.

Summary of the Invention

The present invention relates to a method for detecting donor cfDNA fraction in a sample of a recipient. The present invention further relates to a method for detecting over-represented SNP (s) of donor cfDNA in an identified region. The methods of the present invention may be used to establish the role of early donor DNA dynamics on post-transplantation manipulation, or provide clinical data for determining the likelihood of transplant rejection, in particular, acute rejection, or monitor immunosuppressive therapy in a subject.

In the first place, the present invention relates to a method for detecting donor cfDNA fraction in a sample of a transplant recipient, comprising:

a) extracting DNA from the sample of the transplant recipient, and

b) measuring donor cfDNA fraction based on SNP genotyping.

In some embodiments, extracting DNA from the sample of the transplant recipient comprises extracting genomic DNA from the sample of the recipient pre-transplant.

In some embodiments, cfDNA is extracted from the sample of the recipient post-transplant.

In some embodiments, cfDNA is extracted from the sample of the recipient post-transplant, preferably, during the first four weeks post-transplant, preferably three weeks post-transplant, more two weeks post-transplant. For example, cfDNA is extracted from the sample of the recipient on

day

1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 post-transplant.

In some embodiments, the method comprises a step of preparing a library of the extracted DNA from the sample of the transplant recipient.

In some embodiments, the method comprises a step of preparing a library of the genomic DNA extracted from the sample of the recipient pre-transplant. Preferably, the genomic DNA is fragmented into 100bp-300bp size in length, more preferably, ～250bp size in length.

In some embodiments, the method comprises a step of preparing a library of cfDNA extracted from the sample of the recipient post-transplant. Preferably, cfDNA is 60bp～200bp size in length.

In some embodiments, the method further comprises a step of sequencing DNA of the prepared library. Preferably, sequencing adaptors are ligated to the fragmented DNA or cfDNA.

In some embodiments, in the step of preparing the library, universal amplification is performed for the extracted DNA. Preferably, the amplification product is sequenced by high-throughput sequencing.

In some embodiments, in the step of preparing the library, end repair is performed for cfDNA.

In some embodiments, in the step of preparing the library, dA-tailing process is performed for the fragmented DNA, or cfDNA. Specifically, a non-template dAMP is added onto 3’ end of the fragmented DNA, or cfDNA.

In some embodiments, the method comprises a step of selecting the genome wide SNPs that have a homozygous genotype in the sample of the recipient pre-transplant, and a different genotype in the sample of the recipient post-transplant. SNP is called if the read genotype differs between the sample of the recipient pre-transplant and the sample of the recipient post-transplant.

In some embodiments, the method comprises a step of calculating the minor allele frequency (MAF) for each of the selected SNPs.

The donor cfDNA fraction is an average of all observed values of donor fraction (herein denoted as β) deducted from each selected SNP _i.

In some embodiments, the donor cfDNA fraction is calculated utilizing the following formula 1:

wherein χ is the average of the list of MAFs, which is a mixture of two normal distributions, one centered at 0.5β and the other centered at β, and 0.5β <χ< β, which transforms into:

0.5x ＜ 0.5β,

β ＜ 2x.

The donor genotype on a given SNP _i is also homozygous (i.e., wildtype/wildtype in recipient and mutant/mutant in donor or the other way around) , β _i is equal to MAF; if the donor genotype is heterozygous, β _i will equal two times MAF (i.e., half of the donor fraction equals MAF) .

Outliers beyond the range of (0.5χ, 2χ) is firstly filtered out. In order to separate values from the two different distributions, the list is sorted, and the best separating value point that minimizes the total intra-group variance is found, which can be described as the above objective function formula 1.

This is known as the objective function of k-means clustering when k=2. However, k-means clustering does not guarantee a global optimum (can be trapped into local optima) , while a greedy algorithm can achieve global optima when k=2, with time complexity O (n) . All values in the sorted list before the best separating value point are considered as samples of the 0.5βdistribution, these values are then multiplied by 2. The mean of the resulted new list of values is the predicted β, i.e., the predicted global donor DNA fraction.

The donor cfDNA fraction of greater than 1%indicates that the transplant is undergoing acute rejection, and the donor cfDNA fraction of less than 1%indicates that the transplant is undergoing borderline rejection, undergoing other injury, or stable.

In some embodiments, the method further comprises a step of determining the occurrence or likely occurrence of active rejection of transplantation using the detected donor cfDNA fraction in the sample of the transplant recipient.

In some embodiments, the sample is a blood sample, a urine sample, or a body fluid sample.

In some embodiments, the transplant recipient is a mammal, preferably, a human.

In some embodiments, the transplant recipient received a transplant selected from organ transplant, tissue transplant, and cell transplant.

In some embodiments, the transplant recipient received a transplant selected from, but not limited to, lung transplant, kidney transplant, heart transplant, liver transplant, pancreas transplant, intestinal transplant, stomach transplant, testis transplant, penis transplant, thymus transplant, uterus transplant, bone transplant, bone marrow transplant, tendon transplant, cornea transplant, uterus transplant, ovary transplant, nerve transplant, pancreas islet cell transplant, blood vessel transplant, heart valve transplant, skin transplant, hand transplant, and/or leg transplant.

In the second place, the present invention relates to a method for detecting over-represented donor cfDNA in a sample of a transplant recipient, comprising:

a) extracting DNA from the sample of the transplant recipient,

b) calculating donor cfDNA fraction based on SNP genotyping, and

c) selecting over-represented donor cfDNA.

In some embodiments, genomic DNA is extracted from the sample of the recipient pre-transplant.

In some embodiments, cfDNA is extracted from the sample of the recipient during the first two weeks post-transplant. For example, cfDNA is extracted from the sample of the recipient on

day

1, 4, 7, 10, or 13 post-transplant.

In some embodiments, the method comprises a step of preparing a library of the genomic DNA extracted from the sample of the recipient pre-transplant. Preferably, the genomic DNA is fragmented into 200bp-300bp size in length, more preferably, ～250bp size in length.

In some embodiments, the method comprises a step of preparing a library of cfDNA extracted from the sample of the recipient post-transplant. Preferably, cfDNA is 60bp～200bp size in length. More preferably, cfDNA is ～170bp size in length.

In some embodiments, the step of calculating donor cfDNA fraction based on SNP genotyping comprises splitting chromosome into windows covering 100-1000kb, preferably, 400-600kb, more preferably, 500kb in size, and calculating donor cfDNA fraction utilizing maximum-likelihood-estimation (MLE) .

Preferably, 22 autosomes are splitted into windows (regions) covering 100-1000kb seperately, preferably, 400-600kb, more preferably, 500kb in size.

In some embodiments, the step of calculating donor cfDNA fraction comprises calculating p-value for each window based on assuming normal distribution of the donor cfDNA fraction values, wherein for each SNP (denoted as SNP _i) , p _i-value can be calculated as follows:

for all the SNPs, the total probability p is the products of each p _i:

given that assuming (under ideal conditions) each SNP has only three possible values of wildtype frequency: {0, 0.5, 1} on both the recipient and the donor;

for each SNP (denoted as SNPi) , the number of reads supporting the wildtype genotype ki and the number of reads supporting the mutant genotype qi are known;

if given β the donor DNA fraction, then the wildtype frequency wi of SNPi in the cfDNA sample have 3 ² = 9 possible values: {0, 0.5β, β, 0.5 (1-β) , 0.5, 0.5 (1+β) , 1-β, 1-0.5β, 1} .

For a given β on a given SNPi, wi that gives the biggest pi is found out. Repeat it to all SNPs, the maximum log p for the given β can be calculated. Different choices of β values yield different maximum log p (i.e., log-likelihood) . By iterating through evenly spaced numbers within a specified interval of possible β values, the method can find out the closest approximation of the best β resulting in a maximum log-likelihood.

Preferably, the step of calculating donor cfDNA fraction comprises correcting the median p-values for multiple tests.

In some embodiments, the step of selecting over-represented donor cfDNA comprises selecting over-represented donor cfDNA having an FDR ≤ 0.1 for each transplant recipient.

In some embodiments, the method further comprises a step of selecting an over-represented region.

An over-represented region is defined as having an FDR ≤ 0.1 for each transplant recipient.

In some embodiments, the method further comprises a step of selecting over-represented SNP having an FDR≤ 0.1 for each transplant recipient.

Preferably, the donor fraction β _i for each selected SNP _i is calculated according to formula 1 in the first place.

p-value for each βi assuming normal distribution is calculated and the median p-values for multiple tests are corrected.

In some embodiments, the method further comprises a step of selecting a region enrichment of significantly over-represented SNPs satisfying: 1) contain ≥ 5 significant SNPs, and (2) each significant SNP is less than 0.5M apart from its nearest significant SNPs.

In some embodiments, the over-represented regions are regions associated with graft-host immune responses and/or antimicrobial activities. For example, the over-represented regions are regions associated with HLA-A/B/C, MICA, DDR1, for example chr6: 30782303-31426881, and/or regions overlapped with the family of β defensin genes, e.g. DEFB103/104/105/106, for example, chr7: 75883092-77138957 and chr8: 7237702-7978545.

In some embodiments, the transplant recipient is a mammal, preferably, a human.

In the third place, the present invention relates to a method for determining the likelihood of transplant rejection in a transplant recipient, comprising the steps of the method in the first place, and/or the steps of the method in the second place.

In the fourth place, the present invention relates to a method for predicting prognosis of transplant rejection in a transplant recipient, comprising the steps of the method in the first place, and/or the steps of the method in the second place.

In the fifth place, the present invention relates to a method for diagnosing a transplant within a transplant recipient as undergoing acute rejection, comprising the steps of the method in the first place, and/or the steps of the method in the second place.

In the sixth place, the present invention relates to an apparatus for detecting donor cfDNA fraction in a sample of a transplant recipient, for determining the likelihood of transplant rejection in a transplant recipient, for predicting prognosis of transplant rejection in a transplant recipient, or for diagnosing a transplant within a transplant recipient as undergoing acute rejection, comprising the following modules:

(1) a module for extracting DNA from the sample of the transplant recipient, and

(2) a module for measuring donor cfDNA fraction based on SNP genotyping.

In some embodiments, the apparatus further comprises a module for preparing a library of the extracted DNA from the sample of the transplant recipient.

In some embodiments, the apparatus further comprises a module for preparing a library of the genomic DNA extracted from the sample of the recipient pre-transplant.

In some embodiments, the apparatus further comprises a module for preparing a library of cfDNA extracted from the sample of the recipient post-transplant.

In some embodiments, the apparatus further comprises a module for sequencing DNA of the prepared library.

In some embodiments, the apparatus further comprises a module for universal amplification.

In some embodiments, the apparatus further comprises a module for selecting the genome wide SNPs that have a homozygous genotype in the sample of the recipient pre-transplant, and a different genotype in the sample of the recipient post-transplant.

In the seventh place, the present invention relates to an apparatus for detecting donor cfDNA fraction in a sample of a transplant recipient, for determining the likelihood of transplant rejection in a transplant recipient, for predicting prognosis of transplant rejection in a transplant recipient, or for diagnosing a transplant within a transplant recipient as undergoing acute rejection, comprising the following modules:

(1) a module for extracting DNA from the sample of the transplant recipient;

(2) a module for calculating donor cfDNA fraction based on SNP genotyping; and

(3) a module for selecting over-represented donor cfDNA.

In the eighth place, the present invention relates to use of an agent for detecting a region satisfying one of the following conditions: over-represented region having an FDR ≤ 0.1 for each transplant recipient;

a region enrichment of significantly over-represented SNPs satisfying: 1) contain ≥ 5 significant SNPs; and (2) each significant SNP is less than 0.5M apart from its nearest significant SNPs;

an over-represented region associated with graft-host immune responses and/or antimicrobial activities;

an over-represented region associated with HLA-A/B/C, MICA, DDR1, for example chr6: 30782303-31426881, and/or regions overlapped with the family of β defensin genes, e.g. DEFB103/104/105/106, for example, chr7: 75883092-77138957 and chr8: 7237702-7978545, for determining the likelihood of transplant rejection in a transplant recipient, for predicting prognosis of transplant rejection in a transplant recipient, or for diagnosing a transplant within a transplant recipient as undergoing acute rejection.

In the ninth place, the present invention relates to use of an agent for detecting donor cfDNA fraction in a sample of a transplant recipient for determining the likelihood of transplant rejection in a transplant recipient, for predicting prognosis of transplant rejection in a transplant recipient, or for diagnosing a transplant within a transplant recipient as undergoing acute rejection.

Brief Description of the Drawings

Figure 1A illustrates the study design and data analysis flowchart.

Figure 1B shows that the global donor DNA fraction for each transplant recipient at each time point.

Figure 2A shows shows log p-value vs chromosomal position of genome-wide SNPs for three recipients.

Figure 2B shows the defined significant regions according to the number of significant SNPs.

Figure 3A is a plot diagram showing hits with a p-value ≤ 0.05.

Figure 3B shows graphs of relative coverage of several genes.

Figure 4 shows a circos plot of estimated regional donor DNA fraction on 500kb windows of patient 1.

Figure 5 shows a circos plot of estimated regional donor DNA fraction on 500kb windows of patient 2.

Figure 6 shows a circos plot of estimated regional donor DNA fraction on 500kb windows of patient 3.

Figure 7 shows that the over-representation is not likely to be caused by copy number variations (duplications) in the Dx samples.

Description of Particular Embodiments of the Invention

As for genome-wide SNP-based global donor DNA fraction estimation

Genome-wide SNPs (excluding sex chromosomes) that have a strictly homozygous genotype in the D0 (gDNA) sample and a different genotype in the Dx (cfDNA) samples are firstly selected. To reduce the effects of rare genetic variants on reproducibility, variants with a gnomAD_ALL population frequency lower than 0.3 are filtered [5] . For each of the selected SNPs, the minor allele frequency (MAF) is calculated [6, 7] . The global donor DNA fraction is considered as an average of all observed values of donor fraction (herein denoted as β) deducted from each selected SNP _i. It is assumed that each β _i follows normal distribution. If the donor genotype on a given SNP _i is also homozygous (i.e., wildtype/wildtype in recipient and mutant/mutant in donor or the other way around) , β _i is equal to MAF; if the donor genotype is heterozygous, β _i will equal two times MAF (i.e., half of the donor fraction equals MAF) . Therefore, the list of MAFs is a mixture of two normal distributions, one centred at 0.5β and the other centred at β. Then the average of the list (denoted as x) is calculated and obtained 0.5β < x < β, which transforms into

0.5x ＜ 0.5β,

β ＜ 2x.

Using the above constraints, outliers beyond the range of (0.5x, 2x) are firstly filtered out. Next, to separate values from the two different distributions, the list is sorted, and the best separating value point that minimizes the total intra-group variance is found, which can be described as the following objective function:

As for region-level donor DNA fraction estimation and over-representation analysis

The above genome-wide estimation method is not suitable for calculating donor DNA fraction in a small region (e.g. 500kb) , because there could be too few SNPs that satisfy the ‘strictly homozygous genotype in D0 sample and a different genotype in the Dx samples’ condition within a given small region. Hence for region-level estimation, a method that does not require D0 to be homozygous, which allows more SNP information to be considered, is utilized. This is a simplified variant of a previously published maximum likelihood-based method [8] .

Maximum-likelihood-estimation (MLE) of donor DNA fraction for a small region: assuming (under ideal conditions) each SNP has only three possible values of wildtype frequency: {0, 0.5, 1} on both the recipient and the donor. For each SNP (denoted as SNP _i) , the number of reads supporting the wildtype genotype k _i and the number of reads supporting the mutant genotype q _i are known. If given β the donor DNA fraction, then the wildtype frequency w _i of SNP _i in the cfDNA sample have 3 ² = 9 possible values: {0, 0.5β, β, 0.5 (1-β) , 0.5, 0.5 (1+β) , 1-β, 1-0.5β, 1} . The probability p _i of observing k _i and q _i given wildtype frequency w _i for SNP _i can be calculated as followed:

Then for all the SNPs, the total probability p is the products of each p _i.

For a given β on a given SNP _i, it is find out the w _i that gives the biggest p _i. Repeat it to all SNPs, the maximum log p for the given β can be calculated. Different choices of β values yield different maximum log p (i.e., log-likelihood) . By iterating through evenly spaced numbers within a specified interval of possible β values, the method can find out the closest approximation of the best β resulting in a maximum log-likelihood.

Unlike the global method which uses only SNPs with homozygous genotypes in D0, the regional method starts with all SNPs. To reduce the effects of rare genetic variants on reproducibility while keeping as much as SNPs as possible for the calculations, variants with a gnomAD_ALL population frequency lower than 0.01 are filtered. 22 autosomes are splitted into 500kb windows, and the donor DNA fraction for each region at each time point is estimated. Last window of each chromosome is less than 500kb; windows with no informative SNPs (i.e., differed D0 and Dx genotypes) are removed. For each transplant recipient, a circos plot of estimated regional donor DNA fraction is drawn for all time points (Figures 4-6) . Assuming normal distribution of the regional donor DNA fraction values, p-values for all windows at each time point are calculated, and the median p-values for multiple tests are corrected using Benjamini-Hochberg adjustment. Over-represented regions are defined as having an FDR ≤ 0.1 for each transplant recipient.

As for SNP-level over-representation analysis

From above genome-wide method, donor fraction β _i for each selected SNP _i is estimated . Then a p-value for each β _i assuming normal distribution is calculated, and the median p-values for multiple tests are corrected using Benjamini-Hochberg adjustment. Significantly over-represented SNP is defined as having an FDR ≤ 0.1 for each transplant recipient. A Manhattan plot of median log adjusted p-value of all selected SNPs over the genome (excluding sex chromosomes) is drawn for each recipient using the qqman R package [9] (Figure 2A) . Regions with enrichment of significantly over-represented SNPs is further defined as regions satisfying: (1) contain ≥ 5 significant SNPs, and (2) each significant SNP is less than 0.5M apart from its nearest significant SNPs. The identified enriched regions that overlapped in all recipients are highlighted in grey colour on the Manhattan plots.

As for functional enrichment analysis

Genes within the identified enriched regions that overlapped in all recipients are retrieved using the UCSC table browser [10] . Enrichment of KEGG pathways and GO terms of biological process are identified using the WebGestalt web [11] tool. Hits with a p-value ≤ 0.05 are plotted in Figure 3A.

As for nucleosome foot-printing

Nucleosome foot-printing of genes of interest is performed as previously described [12, 13] . A nucleosome footprint for a gene is the relative coverage (position depth divided by average depth) at ± 1kb of the transcription start site (TSS) . Here the mean of the three D0 samples and mean of the total 15 Dx samples are used as input for plotting, to visualize changes of nucleosome occupancy pre-and post-transplant.

In the genome-wide SNP-based global donor DNA fraction estimation, consistent with previous findings, donor DNA fraction peaked immediately after transplantation (day 1) and fell quickly (by day 4) . Interestingly, after the sharp decrease at day 4, the recipient -patient 3, shows an acute relapse during day 10-13, while patients 1&2 show flattening/slow-decreasing trends. This early dynamics is in line with patient outcome (Table 3) : patient 3 is detected positive of anti-HLA-II antibodies at 26 days post-transplant and developed pleural effusion, a sign of acute lung injury at 5 weeks post-transplant; while patients 1&2 show no signs of allograft dysfunction 18 months post-transplant.

In region-level donor DNA fraction estimation and over-representation analysis, 0.2%(13/5435) of the windows significantly (FDR ≤ 0.1) over-represneted in all three patients is found, mainly distributed at 1p36, 1q21, 9p12-13, 20q11 and 21q11 (table 2) . However, the over-representation was not likely to be caused by copy number variations (duplications) in the Dx samples, because the levels of duplication in D0 samples are higher at these significant regions compared to Dx samples, and the absence of significant regions where higher levels of duplication occur in Dx versus D0 samples.

In SNP-level over-representation analysis, 3 significant regions are identified, namely chr6: 30782303-31426881, chr7: 75883092-77138957 and chr8: 7237702-7978545, which are consistently enriched with over-represented SNPs in all three recipients. Significant regions are defined as having ≥ 5 significant SNPs, each being less than 500kb apart from its closest significant SNPs (Figure 2A and 2B) . Interestingly, chr6: 30782303-31426881 overlapped with the human leukocyte antigen (HLA) region, which includes a series of immune-related genes such as HLA-A/B/C, MICA, DDR1; chr8: 7237702-7978545 overlapped with the family of βdefensin genes, e.g. DEFB103/104/105/106.

All genes within the enriched regions are retrieved, and enrichment analysis is performed to identify significant KEGG pathways and biological processes (p ≤ 0.05; Figure 3A) . The results revealed strong associations with graft-host immune responses and antimicrobial activities.

To further investigate the causes of the “over-shedding” , nucleosome footprints of genes within the significant regions pre-and post-transplant are computed. Characteristics of open chromatin are found in some of the genes post-transplant, suggesting active transcriptions. Among these likely active genes, some are known to be expressed in lung epithelial during inflammation and transplantation, e.g. DEFB103, DDR1 and MICA; some known to be expressed in both the host and the graft, e.g. HLA-B/C. Interestingly, DEFB103 and DDR1, which are expected to be expressed only in the graft, showed different nucleosome footprints pre-and post-transplant, while genes known to express in both, e.g. HLA-B/C, preserved the shape of the footprints.

Actively transcribed regions are considered more likely to have open chromatin structures, which in turn leads to more nuclease-mediated degradation. It is believed that the major source of plasma cfDNA is dying cells resulted from apoptosis/necrosis. A great lymphocyte turnover post-transplant can be inferred from the white blood cell count , indicating the host lymphocytes as the major source of host cfDNA. Relatively high gene expression in these cells can result in the seemingly “over-shedding” phenomenon –it is in fact the “over-degradation” of host cfDNA at these regions, rather than the “over-sheeding” of graft cfDNA. However, for genes more actively or only expressed in the graft, there should be extra sources of cfDNA release, e.g. active secretion via extracellular vesicles, because otherwise they would rather appear “under-shedding” .

The descriptions of particular embodiments and examples are provided by way of illustration and not by way of limitation. Those skilled in the art will readily recognize a variety of noncritical parameters that could be changed or modified to yield essentially similar results.

Examples

Patients and Sample collection

Three patients who have received lung transplantation (LTx) at The First Affiliated Hospital of Guangzhou Medical University are included in this study. Whole blood samples were drawn from each LTx recipient at Day 0 (denoted as D0 sample; pre-transplant) , and

Day

1, 4, 7, 10, 13 (denoted as Dx samples; post-transplant) . Hence for each of the three patients we have a D0 sample for recipient genotyping and five Dx samples containing donor-derived cfDNA (see Figure 1A) .

DNA extraction, library preparation and whole-genome sequencing

Genomic DNA (gDNA) is extracted from D0 samples and cfDNA is extracted from Dx samples. In brief, purified gDNA from D0 samples is extracted using QIAamp DNA Blood Mini Kit (Qiagen, Germen) ; cfDNA from Dx samples is extracted using the QIAamp Circulating Nucleic Acid Kit (Qiagen, Germen) , according to the manufacturer’s instructions, separately.

Library preparation is performed with MGIEasy Universal DNA Library Prep Set and Customer self-developed kit (MGI, Shenzhen, China) . For gDNA, it is fragmented into ～250bp insert size length. A non-templated dAMP is added onto the 3’ end of the fragments, which is known as dA-tailing. As for cfDNA, which is 60-200bp, for example, ～170bp, end repair and dA-tailing processes are performed. Adapter ligation and Pre-PCR are performed, to obtain high-quality libraries before sequencing. DNA quantification is performed with Qubit 3.0 (Thermo Fisher Scientific, USA) . DNB preparation kits and Sequencing kits (MGI, Shenzhen, China) are used to pump DNBs (DNA nanoballs) and load the DNBs onto the Patterned Array chip. Pair-end, 150bp sequencing is performed using MGISEQ-2000 with an expected amount of data >150Gb per sample.

Data pre-processing

Raw sequencing reads are subjected to quality check and filter using Fastp [1] . Read mapping to human genome hg19 is performed using BWA MEM with the command “bwa mem -t 8 -M” [2] . Duplicates are removed and local realignment is performed using the GATK bundle [3] . Genotypes (pileup reads) of the D0 and Dx samples on all positions are determined using the mpileup function of Samtools [4] . SNPs are called if the read genotype (s) differ between D0 and Dx samples (see Figure 1A) .

Example 1 Genome-wide SNP-based global donor DNA fraction estimation

The cfDNA of 15 plasma samples (denoted as Dx samples) from three lung transplant recipients at multiple time points (Day 1/4/7/10/13) during the first two weeks post-transplant, plus their genomic DNA obtained pre-transplant (D0 samples) , using deep (～50X) whole genome sequencing is examined in depth (see Figure 1A) . The global donor DNA fraction for each transplant recipient at each time point is estimated based on genome-wide SNP genotyping [1, 2] (see Figure 1B) . Consistent with previous findings, donor DNA fraction peaks immediately after transplantation (day 1) and fell quickly (by day 4) . Interestingly, after the sharp decrease at day 4, one of the recipient -patient 3, shows an acute relapse during day 10-13, while patients 1&2 show flattening/slow-decreasing trends. This early dynamics is in line with patient outcome (Table 3) : patient 3 is detected positive of anti-HLA-II antibodies at 26 days post-transplant and developed pleural effusion, a sign of acute lung injury at 5 weeks post-transplant; while patients 1&2 show no signs of allograft dysfunction 18 months post-transplant.

Figure 1. A) Study design and data analysis flowchart. Blood samples were collected at day 0 (pre-transplant) ,

day

1, 4, 7, 10, 13 (post-transplant) from recipients of lung transplantation (LTx) . Genomic DNA (gDNA) and cell-free DNA (cfDNA) was extracted from the blood samples and subjected to high coverage whole-genome sequencing. Sequencing data was analysed together with follow-up clinical information. A global donor DNA fraction is estimated for each patient at each time point. Over-representation of graft DNA was examined at SNP-level and region-level. See supplementary Methods for a full description of data processing and analysis. B) Dynamics of global donor DNA fraction during the first two weeks after LTx. Global donor DNA fraction of the three patients at

days

1, 4, 7, 10, 13 (post-transplant) were predicted using the genome-wide SNP-based method. Detailed values of global donor DNA fraction for all samples can be found in table 1.

Predicted global donor DNA fraction for each patient at each time point.

Table 1

Table 3

Example 2 Region-level donor DNA fraction estimation and over-representation analysis

It is firstly examined at a region-level by splitting each chromosome into 500kb-windows and used a maximum-likelihood-based method [3] to estimate a donor DNA fraction for each window (Figures 4-6) . P-value for each window is calculated assuming a normal distribution. It is found 0.2% (13/5435) of the windows significantly (FDR ≤ 0.1) over-represented in all three patients, mainly distributed at 1p36, 1q21, 9p12-13, 20q11 and 21q11 (Table 2) . It is noticed some overlap of these regions with previously reported regions of structural complexity [4] . However, the over-representation is not likely to be caused by copy number variations (duplications) in the Dx samples, because the levels of duplication in D0 samples are higher at these significant regions compared to Dx samples, and the absence of significant regions where higher levels of duplication occurs in Dx versus D0 samples (Figure 7) . Figures 4-6. Circos plots of estimated regional donor DNA fraction on 500kb windows for each recipient at each time point. Time points are arranged from inside to outside -the innermost circle is day 1, and the outermost circle is day 13. The heights of the bars are proportional to the values of the estimated donor DNA fraction. Blank regions indicate lack of informative SNPs at the corresponding window. The circos plot is powered by Bio-oviz circos plot tool: https: //bio. oviz. org/demo-project/analyses/Circos.

Table 2:

500kb-window regions with significant graft DNA over-representation

chrom	start	end	band
chr1	17000001	17500000	1p36.13
chr1	143500001	144000000	1q21.1
chr1	144000001	144500000	1q21.1
chr1	144500001	145000000	1q21.1
chr1	145000001	145500000	1q21.1
chr1	148500001	149000000	1q21.2
chr1	149500001	150000000	1q21.2
chr9	39500001	40000000	9p13.1
chr9	42000001	42500000	9p12
chr9	42500001	43000000	9p12
chr9	43000001	43500000	9p12
chr20	29000001	29500000	20q11.1-21
chr21	14000001	14500000	21q11.1-2

Example 3 SNP-level over-representation analysis

Over-representation at individual SNP-levels is examined. The donor fraction β _i for each selected SNP _i is deduced and p-value for each β _i is calculated. Significantly over-represented SNP is determined as being called in all three patients with FDR ≤ 0.1. Figure 2A shows log p-value (adjusted) vs chromosomal position of genome-wide SNPs for the three recipients. 3 significant regions are identified, namely chr6: 30782303-31426881, chr7: 75883092-77138957 and chr8: 7237702-7978545, which are consistently enriched with over-represented SNPs in all three recipients. Significant regions are defined as having ≥ 5 significant SNPs, each being less than 500kb apart from its closest significant SNPs (Figures 2A and 2B) . Interestingly, chr6: 30782303-31426881 overlaps with the human leukocyte antigen (HLA) region, which includes a series of immune-related genes such as HLA-A/B/C, MICA, DDR1; chr8: 7237702-7978545 overlaps with the family of β defensin genes, e.g. DEFB103/104/105/106.

Example 4 Functional enrichment analysis

Genes within the identified enriched regions that overlapped in all recipients were retrieved using the UCSC table browser [10] . Enrichment of KEGG pathways and GO terms of biological process were identified using the WebGestalt [11] web tool. Hits with a p-value ≤0.05 were plotted in Figure 3A.

All genes within the enriched regions are retrieved, and enrichment analysis is performed [5] to identify significant KEGG pathways and biological processes (p ≤ 0.05; Figure 3A) . The results revealed strong associations with graft-host immune responses and antimicrobial activities.

Example 5 Nucleosome foot-printing

Nucleosome foot-printing of genes of interest were performed as previously described [12, 13] . A nucleosome footprint for a gene is the relative coverage (position depth divided by average depth) at ± 1kb of the transcription start site (TSS) . Here we used the mean of the three D0 samples and mean of the total 15 Dx samples as input for plotting, to visualize changes of nucleosome occupancy pre-and post-transplant.

To further investigate the causes of the “over-shedding” , nucleosome footprints [6] of genes within the significant regions pre-and post-transplant are computed. Characteristics of open chromatin are found in some of the genes post-transplant, suggesting active transcriptions (Figure 3B) . Among these likely active genes, some are known to be expressed in lung epithelial during inflammation and transplantation, e.g. DEFB103, DDR1 and MICA [7–9] ; some known to be expressed in both the host and the graft, e.g. HLA-B/C. Interestingly, DEFB103 and DDR1 which are expected to be expressed only in the graft, showed different nucleosome footprints pre-and post-transplant, while genes known to express in both, e.g. HLA-B/C, preserved the shape of the footprints.

Actively transcribed regions are considered more likely to have open chromatin structures, which in turn leads to more nuclease-mediated degradation [6] . It is believed that the major source of plasma cfDNA is dying cells resulted from apoptosis/necrosis. A great lymphocyte turnover post-transplant can be inferred from the white blood cell count (Table 3) , indicating the host lymphocytes as the major source of host cfDNA. Relatively high gene expression in these cells can result in the seemingly “over-shedding” phenomenon –it is in fact the “over-degradation” of host cfDNA at these regions, rather than the “over-sheeding” of graft cfDNA. However, for genes more actively or only expressed in the graft, there should be extra sources of cfDNA release, e.g. active secretion via extracellular vesicles [10] , because otherwise they would rather appear “under-shedding” .

Claims

A method for detecting donor cfDNA fraction in a sample of a transplant recipient, comprising:

a) extracting DNA from the sample of the transplant recipient, and

b) measuring donor cfDNA fraction based on SNP genotyping.
The method of claim 1, wherein genomic DNA is extracted from the sample of the recipient pre-transplant, and cfDNA is extracted from the sample of the recipient post-transplant, preferably, cfDNA is extracted from the sample of the recipient during the four weeks, preferably three weeks, more preferably two weeks post-transplant.
The method of claim 1, further comprising a step of preparing a library of the extracted DNA from the sample of the transplant recipient, preferably, further comprising a step of preparing a library of the genomic DNA extracted from the sample of the recipient pre-transplant, and a step of preparing a library of cfDNA extracted from the sample of the recipient post-transplant, more preferably, the genomic DNA is fragmented into 200bp-300bp size in length, more preferably, 250 bp size in length.
The method of claim 3, further comprising a step of sequencing DNA of the prepared library, preferably, universal amplification is performed for the extracted DNA, preferably, dA-tailing process is performed for the fragmented DNA, or cfDNA.
The method of claim 1, further comprising a step of selecting the genome wide SNPs that have a homozygous genotype in the sample of the recipient pre-transplant, and a different genotype in the sample of the recipient post-transplant.
The method of claim 1, further comprising a step of calculating the minor allele frequency (MAF) for each of the selected SNPs.
The method of claim 1, wherein the donor cfDNA fraction is an average of all observed values of donor fraction (herein denoted as β) deducted from each selected SNP _i, preferably, the donor cfDNA fraction is calculated utilizing the following formula 1:

wherein, χ is the average of the list of MAFs, which is a mixture of two normal distributions, one centered at 0.5β and the other centered at β, and 0.5β <χ< β, which transforms into:

0.5x ＜ 0.5β,

β ＜ 2x,

if the donor genotype on a given SNP _i is homozygous, β _i is equal to MAF; if the donor genotype is heterozygous, β _i will equal two times MAF; preferably, outliers beyond the range of (0.5χ, 2χ) are filtered out.
The method of claim 7, wherein the donor cfDNA fraction of greater than 1%indicates that the transplant is undergoing acute rejection, and the donor cfDNA fraction of less than 1%indicates that the transplant is undergoing borderline rejection, undergoing other injury, or stable.
A method for detecting over-represented donor cfDNA in a sample of a transplant recipient, comprising:

a) extracting DNA from the sample of the transplant recipient,

b) calculating donor cfDNA fraction based on SNP genotyping, and

c) selecting over-represented donor cfDNA.
The method of claim 9, wherein genomic DNA is extracted from the sample of the recipient pre-transplant, and cfDNA is extracted from the sample of the recipient post-transplant, preferably, cfDNA is extracted from the sample of the recipient during the first two weeks post-transplant.
The method of claim 9, further comprising a step of preparing a library of the extracted DNA from the sample of the transplant recipient, preferably, further comprising a step of preparing a library of the genomic DNA extracted from the sample of the recipient pre-transplant, and a step of preparing a library of cfDNA extracted from the sample of the recipient post-transplant, more preferably, the genomic DNA is fragmented into 200bp-300bp size in length, more preferably, 250 bp size in length.
The method of claim 9, further comprising a step of sequencing DNA of the prepared library, preferably, universal amplification is performed for the extracted DNA, preferably, dA-tailing process is performed for the fragmented DNA, or cfDNA.
The method of claim 9, further comprising a step of calculating donor cfDNA fraction based on SNP genotyping, which comprises splitting chromosome into windows covering 100-1000kb, preferably, 400-600kb, more preferably, 500kb in size, and calculating donor cfDNA fraction utilizing maximum-likelihood-estimation (MLE) , preferably, 22 autosomes are splitted into windows covering 100-1000kb, preferably, 400-600kb in size, more preferably, 500kb in size.
The method of claim 13, wherein the step of calculating donor cfDNA fraction comprises calculating p-value for each window based on assuming normal distribution of the donor cfDNA fraction values,

wherein, for each SNP (denoted as SNP _i) , p _i-value can be calculated as follows:

for all the SNPs, the total probability p is the products of each p _i:
The method of claim 14, wherein for a given β on a given SNPi, wi that gives the biggest pi is found out, repeat it to all SNPs, the maximum log p for the given β is calculated, preferably, by iterating through evenly spaced numbers within a specified interval of possible βvalues, the closest approximation of the best β resulting in a maximum log-likelihood is found out.
The method of claim 14, wherein the step of calculating donor cfDNA fraction comprises correcting the median p-values for multiple tests.
The method of claim 9, wherein the step of selecting over-represented donor cfDNA comprises selecting over-represented donor cfDNA having an FDR ≤ 0.1 for each transplant recipient.
The method of claim 9, further comprising a step of selecting an over-represented region, which is defined as having an FDR ≤ 0.1 for each transplant recipient, and/or a step of selecting over-represented SNP having an FDR≤ 0.1 for each transplant recipient, and/or a step of selecting a region enrichment of significantly over-represented SNPs satisfying: 1) contain ≥ 5 significant SNPs, and (2) each significant SNP is less than 0.5M apart from its nearest significant SNPs, preferably, the over-represented regions are regions associated with graft-host immune responses and/or antimicrobial activities, more preferably, the over-represented regions are regions associated with HLA-A/B/C, MICA, DDR1, for example chr6: 30782303-31426881, and/or regions overlapped with the family of β defensin genes, e.g. DEFB103/104/105/106, for example, chr7: 75883092-77138957 and chr8: 7237702-7978545.
The method of any one of claims 1-18, wherein the sample is a blood sample, a urine sample, or a body fluid sample, and wherein the transplant recipient is a mammal, preferably, a human, preferably, the transplant recipient receives a transplant selected from organ transplant, tissue transplant, and cell transplant, more preferably, the transplant recipient received a transplant selected from, lung transplant, kidney transplant, heart transplant, liver transplant, pancreas transplant, intestinal transplant, stomach transplant, testis transplant, penis transplant, thymus transplant, uterus transplant, bone transplant, bone marrow transplant, tendon transplant, cornea transplant, uterus transplant, ovary transplant, nerve transplant, pancreas islet cell transplant, blood vessel transplant, heart valve transplant, skin transplant, hand transplant, and/or leg transplant.
A method for determining the likelihood of transplant rejection in a transplant recipient, or a method for predicting prognosis of transplant rejection in a transplant recipient, or a method for diagnosing a transplant within a transplant recipient as undergoing acute rejection, comprising the steps of the method of any one of claims 1-8, and/or the steps of the method of any one of claims 9-18.
An apparatus for detecting donor cfDNA fraction in a sample of a transplant recipient, for determining the likelihood of transplant rejection in a transplant recipient, for predicting prognosis of transplant rejection in a transplant recipient, or for diagnosing a transplant within a transplant recipient as undergoing acute rejection, comprising the following modules:

(3) a module for extracting DNA from the sample of the transplant recipient, and

(4) a module for measuring donor cfDNA fraction based on SNP genotyping.
The apparatus of claim 21, further comprising a module for preparing a library of the extracted DNA from the sample of the transplant recipient, and a module for sequencing DNA of the prepared library, preferably, the module for preparing a library of the extracted DNA from the sample of the transplant recipient is a module for preparing a library of the genomic DNA extracted from the sample of the recipient pre-transplant, and a module for preparing a library of cfDNA extracted from the sample of the recipient post-transplant, more preferably, further comprising the apparatus further comprises a module for universal amplification.
The apparatus of claim 21, further comprising a module for selecting the genome wide SNPs that have a homozygous genotype in the sample of the recipient pre-transplant, and a different genotype in the sample of the recipient post-transplant.
An apparatus for detecting donor cfDNA fraction in a sample of a transplant recipient, for determining the likelihood of transplant rejection in a transplant recipient, for predicting prognosis of transplant rejection in a transplant recipient, or for diagnosing a transplant within a transplant recipient as undergoing acute rejection, comprising the following modules:

(1) a module for extracting DNA from the sample of the transplant recipient;

(2) a module for calculating donor cfDNA fraction based on SNP genotyping; and

(3) a module for selecting over-represented donor cfDNA, preferably, calculating donor cfDNA fraction based on SNP genotyping according to anyone of claims 6-8 and 14-18.
The apparatus of claim 24, further comprising a module for preparing a library of the extracted DNA from the sample of the transplant recipient, and a module for sequencing DNA of the prepared library, preferably, the module for preparing a library of the extracted DNA from the sample of the transplant recipient is a module for preparing a library of the genomic DNA extracted from the sample of the recipient pre-transplant, and a module for preparing a library of cfDNA extracted from the sample of the recipient post-transplant, more preferably, further comprising the apparatus further comprises a module for universal amplification.
Use of an agent for detecting a region satisfying one of the following conditions: over-represented region having an FDR ≤ 0.1 for each transplant recipient;

a region enrichment of significantly over-represented SNPs satisfying: 1) contain ≥ 5 significant SNPs; and (2) each significant SNP is less than 0.5M apart from its nearest significant SNPs; an over-represented region associated with graft-host immune responses and/or antimicrobial activities; an over-represented region associated with HLA-A/B/C, MICA, DDR1, for example chr6: 30782303-31426881, and/or regions overlapped with the family of β defensin genes, e.g. DEFB103/104/105/106, for example, chr7: 75883092-77138957 and chr8: 7237702-7978545, for determining the likelihood of transplant rejection in a transplant recipient, for predicting prognosis of transplant rejection in a transplant recipient, or for diagnosing a transplant within a transplant recipient as undergoing acute rejection.
Use of an agent for detecting donor cfDNA fraction in a sample of a transplant recipient for determining the likelihood of transplant rejection in a transplant recipient, for predicting prognosis of transplant rejection in a transplant recipient, or for diagnosing a transplant within a transplant recipient as undergoing acute rejection.