Introduced into clinical practice in 2011, non-invasive prenatal testing (NIPT) allows detection of chromosomal aneuploidies in the fetus using maternal blood samples [1]. Multiple studies have shown that one of the key factors affecting the result of this test is the fetal DNA fraction [2, 3]. The test loses its sensitivity and can come out false-negative if the amount of fetal DNA in the sample is insufficient [3].

The fetal DNA fraction is easy to measure in women carrying a male fetus. This is done by comparing the number of Y reads with the read counts for autosomal chromosomes. In contrast, pregnancy with a female fetus complicates quantification of fetal DNA. Currently existing methods for estimating the proportion of fetal DNA in the total cell-free circulating DNA (cfDNA) are based on the detection and quantification of DNA fragments whose origin can be established. Some of these methods make use of Y-chromosome-specific fragments only found in male fetuses. Others are not sex-based and rely on the analysis of differentially methylated cfDNA fragments [4], SNPs [5–7], unequal sizes of fetal and maternal DNA fragments [8], and distribution of fetal DNA fragments across the genome [9–11].

Targeted sequencing of single nucleotide polymorphisms (SNPs) can be employed to determine the fetal DNA fraction and enables genetic-based identification of the sample. Besides, it can be used in non-invasive paternity and prenatal testing [12, 13].

The aim of this work was to develop a method capable of measuring the fetal DNA fraction regardless of the fetus’s sex using targeted SNP sequencing.

METHODS

Selection of single nucleotide polymorphisms

Seventy-three polymorphisms were selected from the HapMap database, a product of the large-scale population research studies [14] (http://hapmap.ncbi.nlm.nih.gov/). The frequency of their heterozygous variants is 49–51% for the CEU population (Northern and Western Europe) and 45–55% for the African (ASW), Chinese (CHD, CHB) and Japanese (JPT) populations. These polymorphisms are located on chromosomes 1–12 no less than 20 million b.p. apart. For each of them, specific amplification primers were selected. The intended size of PCR products was 110 b.p.

DNA and plasma samples

The frequency of the selected polymorphisms in the Russian population was estimated based on the analysis of 827 DNA samples isolated from donors’ blood. The fetal DNA fraction was measured in 87 plasma samples obtained from 45 women pregnant with a male fetus and 42 women carrying a female fetus.

Frequency of polymorphisms in the Russian population

Because no large-scale population data are available describing the frequency of various polymorphisms in the Russian population and because the polymorphisms we selected represented non-Russian populations, our estimates can differ from the published data. In this work we estimated the frequency of the studied polymorphisms in the Russian population using targeted sequencing of DNA samples pooled at equal concentrations.

Ten pools of 827 samples (51 to 114 samples per pool) were sequenced. Prior to DNA pooling, we determined DNA concentrations in the samples by real-time polymerase chain reaction (PCR). To estimate the frequency of the studied polymorphisms in the Russian population, we added up the frequencies obtained for each sequenced pool, with due account of the number of samples in the pool. The resulting figures were compared with the data generated by the sequencing of 87 individual samples.

Estimation of fetal DNA fraction

The fetal DNA fraction was estimated after sequencing 53 frequent polymorphisms that had been selected based on the preliminary sequence data analysis for pooled samples.

To estimate the fetal DNA fraction, we relied on the polymorphisms for which the frequency of one allele was over 80% but below 99.5%, assuming that the mother had a homozygous genotype and the fetus was heterozygous. The fetal DNA fraction was calculated according to the formula
ff = 2 • B / (A + B), where A is a more abundant and B is a less abundant allele. The fetal DNA fraction was presented as a median of values obtained for all informative polymorphisms. The fetal DNA fraction estimated by SNP genotyping was compared to the value obtained by counting the proportion of DNA molecules originating from the Y chromosome.

Sequencing

Libraries of PCR products were prepared according to the manufacturer’s protocol (Thermo Fisher Scientific Inc., USA). Sequencing was performed on Ion Proton and Ion S5 (Thermo Fisher Scientific Inc., USA) according to the manufacturer’s protocol.

Data analysis

The results were processed using Torrent Server 4.4.3. The sequences were aligned against the reference genome ver. GRCh37/hg19 using TMAP (Thermo Fisher Scientific Inc., USA).Then the reads were counted for each allele located in the genomic regions corresponding to the selected polymorphisms using an original script and the pysam module [15]. Only those fragments were eligible for the analysis for which the alignment quality was >30 and the size was >80% of the expected length.

RESULTS

Sequencing of pooled samples

Data generated by the sequencing of pooled samples are presented in tab. 1. Upon assessing the performance of the method in general and the frequency of the studied polymorphisms, we selected 53 SNPs for further analysis. Infrequent polymorphisms were excluded. tab. 2 shows the frequencies of 53 SNPs in the pooled samples and 87 individual samples. The median difference between the “pooled” and “individual” frequencies was 3.4%.

Results of fetal DNA fraction estimation

The average number of polymorphisms with a homozygous genotype in the mother was 28 (25–32), of them 14 (10–18) were informative. Fig. 1 compares the estimates of the fetal DNA fraction obtained through SNP genotyping and Y chromosome read count; the correlation index is 0.7.

DISCUSSION

We have estimated the frequency of the selected polymorphisms in the studied population using sequencing of pooled samples. We have shown that sequencing of pooled samples and genotyping of individual samples produce comparable results. Targeted sequencing of a small number of frequent polymorphisms is a feasible method for estimating the fetal DNA fraction independent of the fetus’s sex. In our work, the correlation between the results obtained by targeted SNP sequencing and Y chromosome read count was lower than in the published studied that used the comparable number of polymorphisms to measure the fetal DNA fraction [7], probably because the authors of that study used molecular identifiers and counted individual molecules.

CONCLUSIONS

The proposed method can be used to estimate the frequency of alleles in frequent polymorphisms. The method allows both estimation of the fetal DNA fraction and genetic identification of the samples and can be used in non-invasive paternity or prenatal screening if the mutation is passed on by the father and is absent in the mother.