Methods and compositions for analyzing nucleic acid

A method for in silico enrichment of sequence read counts from fetal nucleic acid with reduced experimental bias, comprising:

• a) sequencing test sample nucleic acid by a non-loci-specific massively parallel sequencing process that generates thousands to millions of sequence reads, which test sample nucleic acid comprises circulating cell-free nucleic acid from a test sample from a pregnant female bearing a fetus;
• b) mapping the thousands to millions of sequence reads to portions of a reference genome;
• c) counting the thousands to millions of sequence reads mapped to the portions, which counting generates counts of the thousands to millions of sequence reads mapped to the portions of the reference genome;
• d) measuring the length of nucleic acid fragments from which the thousands to millions of sequence reads are generated;
• e) selecting counts for the thousands to millions of sequence reads from nucleic acid fragments that are shorter than about 160 bases, thereby generating selected counts of the thousands to millions of sequence reads, wherein the selected counts are enriched for counts of reads from fetal nucleic acid;
• and
• f) normalizing the selected counts of the thousands to millions of sequence reads for the test sample according to an experimental bias for the test sample, an experimental bias for each of multiple samples from multiple pregnant females, and counts of sequence reads mapped to each of the genomic portions for the multiple samples, thereby generating normalized selected counts, wherein experimental bias in the normalized selected counts is reduced.