Compositions and methods for accurately identifying mutations

Abstract

The present disclosure provides compositions and methods for accurately detecting mutations by uniquely tagging double stranded nucleic acid molecules with dual cyphers such that sequence data obtained from a sense strand can be linked to sequence data obtained from an anti-sense strand when sequenced, for example, by massively parallel sequencing methods.

1 .- 38 . (canceled) 
     
     
         39 . A method for sequencing double-stranded target nucleic acid molecules, comprising:
 (a) ligating the double-stranded target nucleic acid molecules to double-stranded cyphers at both ends to form cypher-target nucleic acid complexes, wherein the double-stranded cyphers each comprise a random or partially random identifier sequence, and further wherein each target nucleic acid molecule is a DNA molecule comprising a target tag sequence that together with the double-stranded cyphers uniquely labels the double-stranded target nucleic acid molecule;   (b) amplifying each strand of the cypher-target nucleic acid complexes to produce a plurality of cypher-target amplification products from each of a first strand and a complementary second strand of the cypher-target nucleic acid complexes;   (c) sequencing the cypher-target amplification products to produce a plurality of first-strand sequencing reads and a plurality of second-strand sequencing reads, wherein the plurality of first-strand sequencing reads and the plurality of second-strand sequencing reads each comprise an identifier sequence and a target tag sequence;   (d) grouping the first-strand sequencing reads and second-strand sequencing reads based on sequence identity of: the identifier sequence, and the target tag sequence, wherein a group comprises sequencing reads from one of the uniquely labeled double-stranded target nucleic acid molecules; and   (e) comparing the first-strand sequencing reads with the second-strand sequencing reads in each group, and generating an error-corrected sequence for a plurality of the double-stranded target nucleic acid molecules by distinguishing erroneous nucleotides in one strand that lack a matched base change in the complementary strand.   
     
     
         40 . The method of  claim 39 , wherein the double-stranded cyphers comprise random identifier sequences. 
     
     
         41 . The method of  claim 39 , wherein the double-stranded cyphers comprise identifier sequences that are not completely random. 
     
     
         42 . The method of  claim 39 , wherein a target tag sequence comprises nucleotides at an end of a target nucleic acid molecule. 
     
     
         43 . The method of  claim 42 , wherein the end of the target nucleic acid molecule is a sheared end. 
     
     
         44 . The method of  claim 39 , wherein the random or partially-random identifier sequence is double-stranded. 
     
     
         45 . The method of  claim 39 , wherein the identifier sequence is at an end of the double-stranded cypher. 
     
     
         46 . The method of  claim 39 , wherein the target tag sequence is 5 nucleotides to 20 nucleotides in length. 
     
     
         47 . The method of  claim 39 , wherein each of the cyphers are unique. 
     
     
         48 . The method of  claim 39 , wherein the erroneous nucleotides comprise a polymerase error that arose during amplification or sequencing. 
     
     
         49 . The method of  claim 39 , wherein the plurality of target nucleic acid molecules comprise a mutation present at a frequency of 2.1×10 −6  or lower. 
     
     
         50 . The method of  claim 39 , further comprising detecting a cancer biomarker in one of the error-corrected sequences, wherein the cancer biomarker comprises a nucleotide mutation. 
     
     
         51 . The method of  claim 39 , further comprising using the error-corrected sequences to assess cancer response to therapy. 
     
     
         52 . The method of  claim 39 , wherein the sample is derived from a human subject having cancer, and wherein the method further comprises detecting in one of the error-corrected sequences a mutation that confers to the cancer resistance to cancer therapy.