Method and system for polynucleotide synthesis Number:7,164,992 from the United States Patent and Trademark Office (PTO) owispatent

Title: Method and system for polynucleotide synthesis

Abstract: Methods and systems for automated polynucleotide synthesis design are provided. Example embodiments provide an Automated Polynucleotide Synthesis Design System ("APSDS"), which automatically generates a synthesis design for a designated target sequence specification. In one embodiment, the APSDS comprises a synthesis design engine, user interface support, a synthesis rules data repository, and a synthesis data repository. The APSDS automatically generates a synthesis design by receiving a target sequence(s) specification, generating a potential synthesis design, evaluating the potential design against synthesis rules, and when the evaluation indicates that the potential design is not successful according to the synthesis rules, adjusting the design to generate a new potential synthesis design and repeating the process of evaluating and adjusting until a potential synthesis design is found that satisfies the synthesis rules or until no solution is found.

Patent Number: 7,164,992 Issued on 01/16/2007 to Mulligan,   et al.

---

Inventors:	Mulligan; John T. (Seattle, WA), Tabone; John C. (Bothell, WA), Brickner; R. Gregg (Seattle, WA)
Assignee:	Blue Heron Biotechnology, Inc. (Bothell, WA)
Appl. No.:	10/104,986
Filed:	March 22, 2002

---

Current U.S. Class:	702/20 ; 365/94; 700/1
Current International Class:	G06F 19/00 (20060101); G05B 15/00 (20060101); G11C 17/00 (20060101)
Field of Search:	702/19-20 703/11 707/102

---

References Cited [Referenced By]

---

U.S. Patent Documents

	4652639		March 1987		Stabinsky
	5942609		August 1999		Hunkapiller et al.
	6521427		February 2003		Evans
	6670127		December 2003		Evans

Foreign Patent Documents

	WO 98/15567		Apr., 1998		WO
	WO 99/14318		Mar., 1999		WO

Other References

Withers-Martinez et al. PCR-based gene synthesis as an efficient approach for expression of the A+T-rich malaria genome. Protein Engineering, vol. 12, pp. 1113-1120 (1999). cited by examiner . Jerala et al., "Regen: program for designing gene assembly," Nucleic Acids Research 16(5):1759-1766, 1988. cited by other . Libertini et al., "Computer-aided gene design," Protein Engineering 5(8):821-825, 1992. cited by other . Makarova et al., "DIROM'-an interactive system for planning experiments on directed mutagenesis and design of artificial genes," Mol Biol (Mosk) 26(1):93-103, 1992. cited by other.

Primary Examiner: Brusca; John S.
Attorney, Agent or Firm: Seed IP Law Group PLLC

---

Claims

---

The invention claimed is:

1. A method in a computer system for automated generation of a design for synthesizing a designated target double-stranded polynucleotide molecule according to a convergent synthesis technique, comprising: decomposing the designated target molecule into a plurality of potential double-stranded fragments, each fragment having an associated fragment definition and having ends that will potentially ligate with ends of other fragments; automatically determining whether the potential double-stranded fragments satisfy a plurality of synthesis criteria for synthesizing fragments and generating an indication of the determination; when the indication indicates that the potential fragments do not satisfy the synthesis criteria, automatically adjusting the fragment definitions of the plurality of potential fragments thereby adjusting at least one joint of the designated target molecule and repeating the automatically determining whether the potential fragments satisfy the plurality of synthesis criteria, the generating the indication of determination, and the automatically adjusting the fragment definitions until the generated indication indicates that the plurality of potential fragments will synthesize properly to create the designated polynucleotide molecule; and generating a design output that indicates the polynucleotide sequences of the plurality of potential fragments as defined by the adjusted fragment definitions and an ordering for combining the plurality of polynucleotide sequences according to a convergent synthesis technique.

2. The method of claim 1 further comprising using the generated design output to automatically control a polynucleotide synthesizer apparatus.

3. The method of claim 1 wherein a fragment definition comprises, for each strand of the associated sequence fragment, an indication of a position in the designated target molecule, an indication of a nucleotide sequence, and an indication of length.

4. The method of claim 1 wherein the automatically adjusting the fragment definitions comprises adjusting at least one of overhang type, overhang length, joint position between a plurality of fragments, or a sequence of nucleotides in at least one strand of the fragment.

5. The method of claim 4 wherein overhang type is adjusted prior to the sequence.

6. The method of claim 4 wherein overhang type is adjusted prior to joint position.

7. The method of claim 4 wherein the sequence is adjusted prior to overhang type.

8. The method of claim 4 wherein the sequence is adjusted prior to joint position.

9. The method of claim 1 wherein the automatically determining whether the potential fragments satisfy the plurality of synthesis criteria incorporates ends-checking rules to determine the ligatability of a plurality of groups of potential fragments.

10. The method of claim 1 wherein the synthesis criteria specify empirically-based ligation rules.

11. The method of claim 1 wherein the synthesis criteria comprises rules that check for approximate complementarity between base pairs of fragment overhangs of a group of potential fragments.

12. The method of claim 11 wherein the rules determine full matches between the base pairs of the fragment overhangs.

13. The method of claim 11 wherein the rules determine partial matches between the base pairs of the fragment overhangs.

14. The method of claim 13 wherein a partial match between two fragment overhangs occurs when three of four base pairs of the respective fragment overhangs are complementary and at least one of three bases of the fragment overhang of the first fragment is approximately complementary to one of three bases of the fragment overhang of the second fragment.

15. The method of claim 13 wherein a partial match between two fragment overhangs occurs when three of four base pairs of the respective fragment overhangs are complementary.

16. The method of claim 13 wherein a partial match between two fragment overhangs occurs when six of nine base pairs of the respective fragment overhangs are complementary.

17. The method of claim 11 wherein approximate complementarity between base pairs of two fragment overhangs is detected when at least one of an C-T base pair and an G-T base pair is present.

18. The method of claim 11 wherein approximate complementarity between base pairs of two fragment overhangs is detected when a first base is empirically found under synthesis conditions to behave as a recognized mate of a second base but is not a recognized mate, where A and T are recognized mates and G and C are recognized mates, and the first base is present in the first fragment overhang and the second base is present in the second fragment overhang.

19. The method of claim 1 wherein the synthesis criteria are dynamically determined.

20. The method of claim 1 wherein the synthesis criteria comprise ends-checking rules that are derived empirically.

21. The method of claim 1 wherein the synthesis criteria comprise a rule for determining the presence of G or C base in mating positions when the overhangs of two fragments are annealed together.

22. The method of claim 1 wherein the indication of the determination of whether the potential fragments satisfy the synthesis criteria generates a score.

23. The method of claim 22 wherein a separate score is generated each time the fragment definitions are adjusted.

24. A computer-readable memory medium containing instructions for controlling a computer processor, when executed, to automatically generate a design for synthesizing a designated target double-stranded polynucleotide molecule according to a convergent synthesis technique, by performing a method comprising: decomposing the designated target molecule into a plurality of potential double-stranded fragments, each fragment having an associated fragment definition and having ends that will potentially ligate with ends of other fragments; automatically determining whether the potential double-stranded fragments satisfy a plurality of synthesis criteria for synthesizing fragments and generating an indication of the determination; when the indication indicates that the potential fragments do not satisfy the synthesis criteria, automatically adjusting the fragment definitions of the plurality of potential fragments thereby adjusting at least one joint of the designated target molecule and repeating the automatically determining whether the potential fragments satisfy the plurality of synthesis criteria, generating the indication of the determination and the automatically adjusting the fragment definitions until the generated indication indicates that the plurality of potential fragments will synthesize properly to create the designated polynucleotide molecule; and generating a design output that indicates the polynucleotide sequences of the plurality of potential fragments as defined by the adjusted fragment definitions and an ordering for combining the plurality of polynucleotide sequences according to a convergent synthesis technique.

25. The memory medium of claim 24 further comprising using the generated design output to automatically control a polynucleotide synthesizer apparatus.

26. The memory medium of claim 24 wherein a fragment definition comprises, for each strand of the associated sequence fragment, an indication of a position in the designated target molecule, an indication of a nucleotide sequence, and an indication of length.

27. The memory medium of claim 24 wherein the automatically adjusting the fragment definitions comprises adjusting at least one of overhang type, overhang length, joint position between a plurality of fragments, or a sequence of nucleotides in at least one strand of the fragment.

28. The memory medium of claim 24 wherein the automatically determining whether the potential fragments satisfy the plurality of synthesis criteria incorporates rules to determine the ligatability of a plurality of groups of potential fragments.

29. The memory medium of claim 28 wherein the rules comprise at least one of empirically-based ligation rules, rules that check for approximately complementarity between base pairs of fragment overhangs of a group of potential fragments, rules that are dynamically determined, or ends-checking rules that are derived empirically.

30. The memory medium of claim 29 wherein the rules that check for approximate complementarity determine full matches between the base pairs of fragment overhangs.

31. The memory medium of claim 29 wherein the rules that check for approximate complementarity determine partial matches between the base pairs of fragment overhangs.

32. The memory medium of claim 24 wherein the indication of the determination of whether the potential fragments satisfy the synthesis criteria generates a score.

33. A computer system for generating a design for synthesizing a designated target double-stranded polynucleotide molecule according to a convergent synthesis technique, comprising: a synthesis rule data repository that contains a plurality of synthesis criteria for synthesizing fragments according to the convergent synthesis technique; a synthesis design engine that is configured to, decompose the designated target molecule into a plurality of potential double-stranded fragments, each fragment having an associated fragment definition and having ends that will potentially ligate with ends of other fragments; retrieve at least one of the plurality of synthesis criteria from the synthesis rule data repository; automatically determine whether the potential double-stranded fragments satisfy the at least one synthesis criteria and generate an indication of the determination; when the indication indicates that the potential fragments do not satisfy the synthesis criteria, automatically adjust the fragment definitions of the plurality of potential fragments thereby adjusting at least one joint of the designated target molecule and repeat the automatic determination of whether the potential fragments satisfy the synthesis criteria, the generation of the indication of the determination, and the automatic adjustment of the fragment definitions until the generated indication indicates that the plurality of potential fragments will synthesize properly to create the designated polynucleotide molecule; and generate a design output that indicates the polynucleotide sequences of the plurality of potential fragments as defined by the adjusted fragment definitions and an ordering for combining the plurality of polynucleotide sequences according to a convergent synthesis technique; and a synthesis design data repository that receives and stores the generated design output.

34. The computer system of claim 33 further comprising a control system that receives the generated design output from synthesis design data repository and uses generated design output to automatically control a polynucleotide synthesizer apparatus.

35. The computer system of claim 33 wherein a fragment definition comprises, for each strand of the associated sequence fragment, an indication of a position in the designated target molecule, an indication of a nucleotide sequence, and an indication of length.

36. The computer system of claim 33 wherein the automatic adjustment of the fragment definitions comprises adjusting at least one of overhang type, overhang length, joint position between a plurality of fragments, or a sequence of nucleotides in at least one strand of the fragment.

37. The computer system of claim 33 wherein the plurality of synthesis criteria for synthesizing fragments contained in the synthesis rule data repository incorporates rules that can be used to determine the ligatability of a plurality of groups of potential fragments.

38. The computer system of claim 37 wherein the rules comprise at least one of empirically-based ligation rules, rules that check for approximately complementarity between base pairs of fragment overhangs of a group of potential fragments, rules that are dynamically determined, or ends-checking rules that are derived empirically.

39. The computer system of claim 38 wherein the rules that check for approximate complementarity determine full matches between the base pairs of fragment overhangs.

40. The computer system of claim 38 wherein the rules that check for approximate complementarity determine partial matches between the base pairs of fragment overhangs.

41. The computer system of claim 33 wherein the generated indication of the determination of whether the potential fragments satisfy the synthesis criteria generates a score.

---

Description

---

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to methods and systems for synthesizing polynucleotides and, in particular, to methods and systems for automatically determining synthesis designs that minimize incorrect results.

2. Background Information

While it is currently possible to achieve in vitro synthesis of genes and other long polynucleotides, the time and effort required for preparing these polynucleotides causes bottlenecks in current genetic research. With the recent genome mapping achieved by the Human Genome Project, the desire for such polynucleotides has increased dramatically. There is thus an increased desire to synthesize sequences of genes rather than generating sequences using traditional molecular biology methods for modifying or cloning genetic sequences. The efficient and cost-effective synthesis of genes is of paramount interest in projects that can take otherwise take months to set up. Moreover, certain experiments become more feasible, such as those in which a researcher can cost effectively and quickly test multiple variants of a gene to find one that works for the desired purpose. In addition, some gene sequences are far too long to synthesize using traditional techniques.

The following definitions and conventions for referring to gene synthesis terminology are used herein to facilitate description. Additional background information is available in Molecular Biology of the Cell, Third Edition. Alberts, et.al., Garland Publishing, Inc., New York and London, 1994, which is herein incorporated by reference in its entirety. A polynucleotide is formed from a plurality of nucleotides, each nucleotide formed from a phosphate, a sugar (deoxyribose), and an organic base--where the base is selected from the purine bases adenine ("A") and guanine ("G") and the pyrimidine bases thymidine ("T") and cytosine ("C"). Because the sugar is deoxyribose, the nucleotides may be referred to deoxyribonucleotides, and the polynucleotide may be referred to as a deoxyribonucleic acid molecule, or DNA. A DNA molecule may be single-stranded or may hybridize to (also referred to as "anneal to" or "anneal with") another DNA molecule to form a double-stranded structure, where the double-stranded structure may also be referred to as a DNA molecule.

The term "gene" as used herein refers to a double-stranded DNA molecule, where the two strands may be distinguished from one another by referring to one strand as the upper or top strand and the other strand as the lower or bottom strand. Thus, either of the terms "gene" or "DNA molecule" may be used to denote any long, double-stranded polynucleotide, where a long polynucleotide contains at least 100 nucleotides. In simplified explanation, gene synthesis is the production of double-stranded DNA molecules from, typically, a text specification of a single-stranded target sequence. Each single-stranded DNA molecule (and thus nucleotide sequence) has two ends (termini), and is associated with a direction (or "handedness"). One of the two termini is referred to as the 3' end while the other termini is referred to as the 5' end. A terminal hydroxyl group of a sugar is normally located at the 3' end of the molecule, and a terminal phosphate group is normally located at the 5' end of the molecule. A single strand sequence of DNA is also referred to as a "polynucleotide sequence" or, if the sequence is less than 100 nucleotides, then the polynucleotide sequence may optionally be referred to as an "oligonucleotide" or "oligo." A textual representation of a 25-base pair double-stranded DNA molecule is written in this description as: 5' CCTGAGAGGACAGTCAATCACAGGA 3' (top strand) (SEQ ID NO:1) 3' GGACTGTCCTGTCAGTTAGTGTCCT 5' (bottom strand) (SEQ ID NO:2)

Under appropriate conditions, a base in a single-stranded DNA sequence typically forms hydrogen bonds to a second base in another single-stranded DNA sequence when the pair of bases (one from each strand) are complements of each other. (Specifically, an "A" bonds with a "T" and a "C" bonds with a "G.") Thus a single strand of DNA binds to another strand of DNA to form a double-stranded DNA molecule when each pair of bases of the sequences (one from each strand) is exactly complementary. A double-stranded sequence of DNA is known as a "fragment" or "DNA fragment." Thus, a DNA fragment is a duplex polymer formed from deoxyribonucleic acid molecules. A DNA fragment may be entirely double-stranded, or may contain both single-stranded and double-stranded regions. The DNA fragments may be, for example, portions of a gene or fragments that when joined form a larger portion of the gene or the entire gene.

Hybridization (annealing) refers to this process of joining of oligos or polynucleotides (both as single-stranded DNA sequences) to generate fragments (double-stranded DNA molecules), and is written herein as: 5' GGACAGTCAA 3'+5' TTGACTGTCC 3'.fwdarw.(SEQ ID NO:3) (SEQ ID NO:4) 5' GGACAGTCAA 3' (SEQ ID NO:3) 3' CCTGTCAGTT 5' (SEQ ID NO:4) Typically, hybridization of single-stranded DNA sequences occurs at low temperature conditions. As the temperature is raised, the double-stranded structure "melts" to form two non-hybridized single-stranded structures.

Once annealed, the resulting double-stranded DNA sequence (or molecule) may have an overhang on one (left or right) or on both sides. For example, a double-stranded DNA sequence with a 5' overhang on the left can be denoted as: 5'CCTGAGAGGACAGTCAATCACAGGA 3' (SEQ ID NO:5) 3' TGTCCTGTCAGTTAGTGTCCT 5' (SEQ ID NO:6) Or, for example, a double-stranded DNA sequence with a 5' overhang on both the left and right can be denoted as: 5'CCTGAGAGGACAGTCAATCAC 3' (SEQ ID NO:7) 3' TGTCCTGTCAGTTAGTGTCCT 5' (SEQ ID NO:6) Other double-stranded DNA sequences may be formed with 3' overhangs or combinations of 3' and 5' overhangs, or even with no overhangs in which case the ends are referred to as "blunt" ends. The overhang length refers to the number of bases that form the particular overhang. So, for example, in either DNA sequence represented above, the overhang length of the left end is 4 (bases).

DNA fragments can be further combined through a process referred to as ligation to form longer length fragments, which can ultimately constitute a gene. Ligation is a higher temperature process whereby, with the assistance of an enzyme (referred to as a ligase), DNA fragments in solution are covalently joined together. The presence of overhanging ends facilitates this process, and ends that are reverse complements of each other tend to bond. For example, fragments: 5' CCTGAGAGGAC 3' (SEQ ID NO:8) 3' TCTCCTGTCAG 5' (SEQ ID NO:9) and 5' AGTCAATCAC 3' (SEQ ID NO:10) 3' TTAGTGGGAC 5' (SEQ ID NO:11) can be ligated to produce the new fragment: 5'CCTGAGAGGACAGTCAATCAC 3' (SEQ ID NO:7) 3' TCTCCTGTCAGTTAGTGGGAC 5' (SEQ ID NO:12)

According to the standard method of gene synthesis, which may be referred to as the shotgun method of gene synthesis, multiple single-stranded polynucleotide sequences are mixed together in order that they can undergo hybridization to form fragments of (double-stranded) sequences. These fragments have overhangs that are designed to be exactly complementary to overhangs present in other fragments, where the exactly complementary overhangs direct two fragments to hybridize together. The hybridized fragments are then exposed to a ligase to achieve ligation between the adjacent fragments. All of this occurs in a single biochemical reaction. Since many fragments are combined in a single ligation reaction, errors are common and may result in incorrect or undesired synthetic gene sequences.

BRIEF SUMMARY OF THE INVENTION

In one aspect, embodiments of the present invention provide computer- and network-based methods and systems for automatically generating a polynucleotide synthesis design for the synthesis of one or more target double-stranded polynucleotide molecules (deoxyribonucleic acid, or "DNA" molecules). In one embodiment, the techniques of the present invention are used to generate designs for the convergent synthesis of polynucleotide sequences, which are optimized to minimize synthesis errors that may occur through undesired ligations. Convergent synthesis is used to create a gene by ligating double-stranded fragments in a (group-wise) hierarchical order so as to minimize incorrect joining of fragments. A polynucleotide synthesis design specifies what oligos and fragments should be generated, and in what order they may be ligated, in order to achieve a desired target sequence(s). In another embodiment, the techniques of the methods and systems of the present invention are extended and modified to include automatically generating synthesis designs for use with other types of synthesis such as solid phase synthesis, shotgun synthesis, and gapped synthesis.

The automated synthesis design process receives a target (designated) polynucleotide sequence specification, decomposes (deconstructs) the target sequence specification, and, using a set of synthesis heuristics (criteria), generates a synthesis design that indicates the set of oligos (and fragments) that need to be synthesized and the order in which they should be ligated to correctly generate the target sequence(s). In one such embodiment, the synthesis design process automatically designs, whenever possible, a group of fragments that can be later synthesized to create a target sequence such that no two incorrect fragment ends can join in a ligation reaction. In some embodiments, different potential synthesis designs are "scored" to indicate a level of success.

In one embodiment, the design process decomposes the target sequence into a series of double-stranded potential fragments, automatically determines whether the potential fragments meet a synthesis criteria, automatically adjusts the potential fragments until they meet the synthesis criteria or until no more adjustments are possible, and outputs a synthesis design that indicates the adjusted potential fragments as fragments that can be used to synthesize a double-stranded polynucleotide molecule defined by the target sequence.

In one embodiment, the synthesis criteria are rules that assist in determining whether a particular synthesis step will be successful. In one such embodiment, the synthesis rules are ligation rules that check, based upon ends of fragments to be ligated, whether a ligation will occur. In another such embodiment, the synthesis rules are derived empirically. In yet another such embodiment, they are dependent upon the synthesis technique being used. In another such embodiment, they are independent of the synthesis technique. In one embodiment, the synthesis rules check for full matches of fragment overhangs. In another embodiment, the synthesis rules check for partial matches of fragment overhangs. In one such embodiment, the partial match detects whether 3 of 4 base pairs are complementarity. In another such embodiment, the partial match detects whether 6 of 9 base pairs are complementary. In one embodiment, the synthesis rules check for the presence of GC combination base pairs.

In some embodiments, the synthesis criteria determine whether fragment ends will properly ligate by checking for approximate complementarity as well as exact complementarity. In one such embodiment, a base pair is approximate complementarity if it is a "C/T" or "G/T" base pair. In another embodiment, an end is approximately complementary to another end, if the ends behave as though they were exactly complementary. In these embodiments, a full match of base pairs of fragment overhangs occurs when all of the base pairs are either exactly complementary or approximately complementary. Also, a partial match of base pairs of fragment overhangs occurs when a portion of all of the base pairs are either exactly complementary or approximately complementary.

In one embodiment, an Automated Polynucleotide Synthesis Design System ("APSDS") is provided. The APSDS comprises a synthesis design engine, user interface support, a synthesis rules data repository, and a synthesis data repository. The synthesis design engine automatically generates a synthesis design by receiving a target sequence(s) specification, generating a potential synthesis design, evaluating the potential design against synthesis rules, and when the evaluation indicates that the potential design is not successful according to the synthesis rules, adjusting the design to generate a new potential synthesis design and repeating the process of evaluating and adjusting until a potential synthesis design is found that satisfies the synthesis rules or until no solution is found.

In one embodiment, adjusting the design comprises changing fragment specifications, so as to adjust attributes of a plurality of potential joints. Parameters such as a desired default oligo length, joint position, overhang type, and overhang length can be varied and adjusted to change the specification of the fragments (the potential design), hence the joint attributes.

In one embodiment, the synthesis rules are coded into the design engine. In another embodiment, the synthesis rules are retrieved from a data repository. In one such embodiment, the synthesis rules are dynamically configured.

In one such embodiment, the synthesis design engine directly or indirectly controls instruments that carry out the synthesis process. The control may be accomplished by generating instrument-specific commands automatically as part of the design. In other embodiments, the control is accomplished by communicating to another software module, such as a scheduling and synthesis control module.

In one embodiment, the synthesis design comprises a build map, which describes the fragments to be synthesized and the order that the fragments should be ligated to produce the target sequence(s). In another embodiment, the synthesis design is generated according to a particular synthesis technique. In one embodiment, the APSDS automatically generates a synthesis design for use with convergent synthesis techniques, which ligates groups of designed fragments in a hierarchical order. In another embodiment, the generated design is used to control the synthesis process of some or all of the target sequence(s).

In one embodiment, configurable parameters can be input to the synthesis design process, such as a specification of a desired target DNA sequence and/or control parameters that affect the synthesis design process, such as desired oligo length. In one such embodiment, the user interface support provides a secure web-based client interface over, for example, a network such as the Internet.

In another aspect, the present invention provides a method of gene synthesis wherein a plurality of distinct relatively small DNA fragments, e.g., two DNA fragments, are combined together under ligation conditions to provide a (preferably one) relatively large DNA fragment sequence (of course, there will be many DNA fragments of that particular sequence). In one embodiment, this plurality is two, while in another embodiment this plurality is three. Subsequently, a plurality of distinct relatively large DNA fragments, each as prepared above, is combined together under ligation conditions to provide a relatively larger DNA fragment. The process of combining selected smaller DNA fragments together to make a larger DNA fragments, and repeating this with different small DNA fragments to make another larger DNA fragment, and then combining selected larger DNA fragments together to make a set of still larger DNA fragments can be repeated as needed to prepare the desired gene. This process can be referred to as convergent approach to gene synthesis, and is illustrated in FIG. 3 wherein 9 "smaller" DNA fragments are, in separate reaction vessels, combined into 5 "larger" DNA fragments, 2 of these 5 larger DNA fragments are combined into 1 "still larger" DNA fragment, this 1 still larger DNA fragment and the remaining 3 larger DNA fragments are, in two reaction vessels, combined to form 2 largest DNA fragments, and these 2 largest DNA fragments are combined to form the gene. This process is also illustrated in FIGS. 6A 6K.

In a convergent synthesis approach, the final gene is formed by combining at least two DNA fragments, where at least two of these "at least two" DNA fragments where themselves formed by combining at least two DNA fragments. This is to be contrast with a serial approach to gene synthesis, where a first DNA fragment is ligated to a second DNA fragment to provide a first ligation product, the first ligation product is ligated to a third DNA fragment to provide a second ligation product, and the second ligation product is ligated to a fourth DNA fragment to form the final gene.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is an example block diagram of an example embodiment of an Automated Polynucleotide Synthesis Design System.

FIG. 2 is an example block diagram of a gene synthesis process that incorporates an example Automated Polynucleotide Synthesis Design System of the present invention.

FIG. 3 is an example schematic diagram of convergent synthesis technique.

FIG. 4 is an example illustration of an interference problem caused by ligating a pair of fragments.

FIG. 5 is an overview flow diagram of a design process used by an example Automated Polynucleotide Synthesis Design System, which incorporates heuristics or other criteria to determine a set of fragments to synthesize independent of the synthesis technique used.

FIGS. 6A 6K are example schematic diagrams of the fragments tested and adjusted at each level in an example APSDS used to generate a convergent synthesis design.

FIG. 7 is a block diagram of a general-purpose computer system for practicing embodiments of the Automated Polynucleotide Synthesis Design System.

FIG. 8 is an example overview flow diagram of steps executed by an example embodiment of an automated synthesis design engine to generate a synthesis design.

FIG. 9 is an example flow diagram of a routine for decomposing a sequence into a synthesis design.

FIG. 10 is an example schematic for determining fragment set definitions for an example set of initial potential fragments.

FIG. 11 is an example flow diagram of a routine for processing fragment sets.

FIG. 12 is an example flow diagram of a routine for evaluating and adjusting the design of fragments referred to by a fragment set.

FIG. 13 is an example schematic of joint position adjustment.

FIG. 14 is an example schematic of joint overhang type adjustment.

FIG. 15 is an example flow diagram for evaluating/scoring a fragment set.

FIG. 16 is an example flow diagram of example synthesis rules used to evaluate ends of a pair of DNA fragments.

DETAILED DESCRIPTION OF THE INVENTION

Embodiments of the present invention provide computer- and network-based methods and systems for automatically generating a polynucleotide synthesis design for the synthesis of one or more target double-stranded polynucleotide sequences (deoxyribonucleic acid, or "DNA" molecules). In one embodiment, the techniques of the present invention are used to generate designs for the convergent synthesis of polynucleotide sequences, which are optimized to minimize synthesis errors that may occur through undesired ligations. Convergent synthesis, as will be described in detail below, is used to create a gene by ligating double-stranded fragments in a (pair-wise) hierarchical order so as to minimize incorrect joining of fragments. As used herein, a polynucleotide synthesis design specifies what oligos and fragments should be generated, and in what order they may be ligated, in order to achieve a desired target sequence(s). Although discussed primarily with respect to convergent synthesis, one skilled in the art will recognize that the techniques of the methods and systems described herein can be extended and modified to include automatically generating synthesis designs for use with other types of synthesis such as solid phase synthesis, shotgun synthesis, and gapped synthesis.

Example embodiments provide automated synthesis design of target sequences through an Automated Polynucleotide Synthesis Design System ("APSDS"). The APSDS takes as input a target (designated) polynucleotide sequence specification, decomposes (deconstructs) the target sequence specification, and, using a set of heuristics and criteria, generates the set of oligos (and fragments) that need to be synthesized and the order in which they should be ligated to correctly generate the target sequence(s). The generated design can then be used to control the synthesis process of some or all of the target sequence(s).

FIG. 1 is an example block diagram of an example embodiment of an Automated Polynucleotide Synthesis Design System. The example APSDS 100 comprises user interface support 102, a synthesis design engine 103, a synthesis rule data repository 105, and a synthesis data repository 104. Through the user interface support 102, the APSDS communicates with clients 101 to specify configurable parameters to the synthesis design process, such as a specification of a desired target DNA sequence and/or control parameters that affect the synthesis design process, such as desired oligo length. In one embodiment, the user interface support 102 provides a secure web-based client interface over, for example, a network such as the Internet. One skilled in the art will recognize that embodiments of the present invention are intended to work with any physical configuration, distributed, local, wireless, or otherwise, that could host one or more parts of the APSDS. The synthesis design engine 103, as will be explained in further detail below, receives a target DNA sequence specification from the user interface support 102 and, using the rules stored in the synthesis rules data repository 105, generates a synthesis design, which is then stored in the synthesis data repository 104. The synthesis design comprises sufficient data to drive a synthesis process using a particular ligation technique. For example, the design may define the set of oligos (and fragments) that need to be synthesized and the order in which they should be ligated to correctly generate the target sequence(s). The synthesis rules contained in the synthesis rules data repository 105 that are used to drive the synthesis design engine 103 may or may not be dependent upon the type of synthesis being used. For example, the criteria used to examine a potential synthesis design to determine whether correct or incorrect ligations will occur are stored in repository 105. In addition, criteria that are only applicable when using particular types of synthesis techniques may be also stored in repository 105. In one embodiment, the synthesis design engine 103 directly or indirectly controls instruments 107, such as oligo synthesizers that carry out the synthesis process, through instrument control systems 106. This may be accomplished, for example, by generating the appropriate instrument-specific commands automatically as part of the design, or under a separate user interface, or by communicating through another software module, such as a scheduling and synthesis control module, potentially provided by another supplier.

FIG. 2 is an example block diagram of a gene synthesis process that incorporates an example Automated Polynucleotide Synthesis Design System of the present invention. In FIG. 2, a specification of a target sequence(s) for which a synthesis design is to be generated is supplied to the APSDS 201. The APSDS 201 automatically generates a synthesis design according to the techniques of the present invention, which are further described below. Part of the generated design is a specification of the oligos that need to be synthesized. These oligo definitions are fed to a DNA sequence synthesizer 202, for example a Beckman Oligo 1000M synthesizer, which synthesizes oligos. The oligos are then placed into storage and/or reaction vessels, e.g., 96-well plates. The synthesized oligos are then subjected to a hybridization process 203, whereby selected oligos are combined into pairs and allowed to form hybrids, i.e., form the double-stranded DNA fragments indicated by the synthesis design that was generated by the APSDS 201. These double-stranded DNA fragments are then subjected to a ligation process 204, whereby DNA fragments are combined in the presence of a ligase in an order specified by the synthesis design that was generated by the APSDS 201 to generate a desired DNA fragment that satisfies the target sequence(s). The order of ligation reactions is dependent upon the synthesis (ligation) technique (scheme) being used and the ligation process 204 supports ligations of pairs or groups of fragments ligated in a hierarchical or iterative fashion to create longer and longer fragments. The ligation process 204 is repeated as needed to provide the target double-stranded sequence. Once the target double-stranded sequence has been synthesized, the sequence is cloned by a cloning process 205 to produce colonies of the target sequence in bacteria. Samples of these colonies are then checked, for example by DNA sequencing, by the quality control testing process 206 to insure correct sequences have been produced. Once verified, the synthesized sequences (the genes) are delivered to the requesting client.

When the gene synthesis procedure of FIG. 2 uses a convergent synthesis process, then the ligation process 204 progressively generates larger and larger DNA fragments from pairs of smaller fragments until the target sequence(s) is generated. FIG. 3 is an example schematic diagram of the convergent synthesis technique. In FIG. 3, multiple levels 301 of combining double-stranded fragments 302 (a hierarchy of levels) are shown as Levels A E. Pairs of fragments on a particular level are ligated to produce the fragments of the next higher level. Specifically, pairs of fragments from Level A are ligated to produce the fragments needed for Level B. Pairs of fragments from Level B are then ligated to produce the fragments needed in the next higher level, and so on, until there are no more fragments to combine. (See, for example, Level E.) In one embodiment, the initial fragments to be used (the fragments of Level A) are determined by dividing the target sequence into polynucleotide sequences of a designated oligo length. Other initial configurations are possible.

In practice, two complications occur with convergent synthesis. First, whenever the number of initial fragments is not a power of two, then, at some level, there will be generated an odd number of fragments. For example, in FIG. 3, there are 5 fragments, labeled with numbers 0 4, created at Level B. One skilled in the art will recognize that several techniques can be used to handle the "extra" fragment. In one such technique, an additional level (see, for example, Level BX) is added to pair-wise ligate the two rightmost fragments first (e.g., fragments 3 and 4) and then that product fragment is used as the rightmost fragment in the next level (see, e.g., fragment 3 of Level C). Another technique is to consolidate all extra fragment ligations on a single level so that successive ligation levels produce only a number of fragments that is a power of two. Other similar techniques and variations thereof are also possible.

Second, for every pair of DNA fragments involved in a ligation reaction, the synthesis process needs to insure that each fragment combines (ligates) properly with the other fragment so as to produce only correct ligations. Since each fragment potentially has two (overhanging) ends, the synthesis process needs to insure that each end of the two ends will only ligate with the desired end of another fragment. In the laboratory, since large numbers of DNA molecules participate in each ligation reaction, all possible combinations of ends will potentially ligate. This is termed the "interference problem," since undesired ligations can interfere with synthesis of a correct sequence. Because each pair of fragments has 4 ends, these ends can potentially combine in 16 (10 different) ways as illustrated in FIG. 4, discussed below. When any two such ends are exactly complementary, they will tend to ligate (other conditions being equal), whether such ligation is desired or not. Thus, the design process needs to insure that only one combination of the four ends will combine.

For example, given two shorter DNA fragments:

##EQU00001## the synthesis process needs to insure that the fragments will ligate only in one way to produce the new desired fragment:

.times. ##EQU00002## The base sequences of the four overhanging ends of the pair of shorter DNA fragments are denoted by an "A," "B," Bp," and "C." The first fragment has end sequences denoted by an "A" (top) and a "B" (bottom), and the second fragment has end sequences denoted by a "Bp" (top) and "C" (bottom). (The second fragment is written 5' to 3' per convention and thus the top and bottom strand are reversed.) The "Bp" sequence of the left end of the second fragment indicates the fact that the overhang of the left end of the second fragment is the exact complement of the overhang of the right end of the first fragment (reverse complement when written 5' to 3', as per convention). In this case, the desired reaction ligates the "B" and "Bp" ends to form a longer fragment with end sequences denoted by the "A" and the "C."

FIG. 4 is an example illustration of an interference problem caused by ligating a pair of fragments. FIG. 4 illustrates 16 possible ligations that can occur with the pair of fragments shown above, only one of which is desired, the combination of ends "B" and "Bp." In the first, second, fifth, and sixth reactions, a first fragment incorrectly ligates with a copy of itself. Similarly, in the eleventh, twelfth, fifteenth, and sixteenth reactions, a second fragment incorrectly ligates with a copy of itself. The other reactions represent ligations of the first and second fragments that are undesired with the exception of the seventh reaction:

.times..times..times. ##EQU00003##

For correct ligation to be possible, the overhang types (5' or 3') must be the same; the overhang lengths must be the same; and the overhangs in the top and bottom strands must be exactly complementary. To complicate matters, empirical evidence used with the present invention has demonstrated that ligations can occur even when the ends of the two fragments are not exactly complementary and even when the overhang lengths are not the same. Specifically, there are instances in the laboratory wherein two ends behave as though they are exactly complementary. This behavior is referred to herein as "approximately complementary;" that is, an end is approximately (or effectively) complementary to another end if it behaves as though it is exactly complementary in a ligation reaction.

Laboratory experiments have shown that certain conditions will tend to generate approximate complementarity. One skilled in the art will recognize that other conditions may be demonstrated empirically, and that such conditions can easily be incorporated into synthesis rules used to determine whether two ends will ligate correctly. However, as described above in connection with FIG. 4, for a ligation composed of two fragments, there are 10 potential ligation products. The only correct product is ligation reaction 7. Ends that are designed using strict Watson Crick complementarity where A binds T and C binds G are insufficient to predict incorrect ligations. In order to discover non-Watson-Crick base pair interactions that result in non-specific ligations we have tested thousands of different combinations of ends have been tested and these results translated into rules for an algorithm used to design gene synthesis strategies.

The nine incorrect ligation reactions shown in FIG. 4 can be divided into two types based on the type of products that are formed: ligation reactions that form incorrect products that are equal in length to two fragments (see the type 1, 6, 11 and 16); and ligation reactions that form incorrect products that are polymers whose lengths are multiples of 3 or more times the length of the fragments (see the types 2/5, 3/9, 4/13, 8/14 and 12/15). All the products, including the correct product, are distinguishable using ligation reactions containing both fragments (normal ligation) or ligation reactions with one fragment (self ligations) with length analysis by polyacrylamide gel electrophoresis (PAGE).

It is desirable to perform an analysis to identify fragments that form incorrect products that are equal in length to two fragments. The test for this type of incorrect ligation is to set up two ligation reactions with only one fragment present in each ligation (self ligation test). There are three potential outcomes from this type of ligation experiment. (1) If the ends are specific, as is the case for ligation reaction 7/10, or non-specific as is the case for the ends of the type similar to 8/14, 4/13 or 3/9, then no ligation product will result. (2) If the ends are of the type similar to ligation reactions 1, 6, 11 and 16, then a ligation product equal in length to two fragments will result. (3) If the ends are of the type similar to ligation reactions 2/5 or 12/15, then any products are polymers whose lengths are multiples of 3 or more time the length of the fragments.

In is also desirable to perform an analysis to identify fragments that form incorrect ligations composed of multiple fragments. The test for this type of incorrect ligation is to set up a standard ligation reaction with two fragments. There are two potential outcomes for this ligation. (1) If the ends are specific, as is for ligation reaction 4/13, or non-specific as is for the ends of the type similar to ligation reactions 6, 1, or 16, then a single product will result. (2) If the ends are of the type similar to ligation reactions 2/5, 8/14, 7/10, 3/9 or 12/15, then multiple products will result.

From the two self ligation and single normal ligation experiments described above, ends can be classified as follows:

Specific ends: If the self ligation reactions have no ligation product and the normal ligation reaction has a single ligation product, all four ends involved are specific with respect to one another. If a self ligation reaction has no ligation products then the ends of the fragment are specific with respect to one another.

Non-specific ends: If the self ligation has no ligation products and the normal ligation has multiple products the end(s) in one fragment interact with end(s) in the other fragment. If the self ligation reaction has one ligation product and the normal ligation has a single ligation product the fragment has one end that is non-specific and reacts with itself. If both self ligation reactions have one ligation product and the normal ligation reaction has one ligation product then the two ends that join the fragments are non-specific with themselves. If the self ligation reaction has multiple ligation products the joints in the fragment are interacting with themselves and/or each other.

Thousands of ends have been evaluated using the strategy described above. The results of these experiments have been compiled and stored in a data repository. The specificity of each end with itself and other ends is recorded. This information is analyzed to build the synthesis (ends-checking) rules that can be incorporated into an APSDS of the present invention.

The APSDS provides an automated solution to the interference problem by automatically designing, whenever possible, a group of fragments that can be later synthesized to create a target sequence such that no two incorrect fragment ends can join in a ligation reaction. Specifically, the APSDS generates a ligation scheme (a specific ordering of fragments and ligation reactions) that insures, wherever possible, that only desired ligations will succeed. In some embodiments, different designs are "scored" to indicate a level of success. Such indications are potentially useful in a laboratory setting in which tradeoffs can be made, for example, in terms of efficiency versus correctness. In summary, the APSDS examines an initial design of fragments in terms of the order in which they will be ligated, determines whether they will generate any undesired ligations (scores the design), and readjusts the fragment design and scores each readjusted design until no undesired ligations are detected.

FIG. 5 is an overview flow diagram of a design process used by an example Automated Polynucleotide Synthesis Design System, which incorporates heuristics or other criteria to determine a set of fragments to synthesize independent of the synthesis technique used. In step 501, a specification of the target sequence is received by the APSDS. In step 502, the APSDS creates an initial potential synthesis design. In one embodiment, this step is performed by decomposing the target sequence (in its double-stranded form) into fragments of a length equal to a desired default oligo length. In step 503, the potential design is evaluated according to heuristics or other criteria. Such criteria may test whether a potential design minimizes the occurrence of incorrect ligations that may result from the fragments specified in the potential design. In one embodiment, the evaluation criteria determine whether the ends involved in each potential ligation reaction will ligate as desired. These ligation rules are preferably empirically based upon results determined in a laboratory. In some embodiments, the evaluation criteria determine whether fragment ends will properly ligate by checking for approximate complementarity as well as exact complementarity. In addition, new or modified evaluation criteria preferably can be dynamically added to the system. In other embodiments, the criteria may be used to evaluate aspects of a potential design other than those that are ligation related. In addition, the evaluation process may provide an indication, such as a score, of the relative success of the potential design. One skilled in the art will recognize that many variations are possible. In step 504, the APSDS determines whether the potential design meets the specified criteria, and, if so, continues in step 505, else continues in step 506. In step 505, the (now successful) potential design is output, and the process ends. Alternatively, in step 506, if the potential design does not meet the criteria, then the design is adjusted to create a new potential design, and then reevaluated in step 503. In one embodiment, parameters such as a desired default oligo length, joint position, overhang type, and overhang length can be varied and adjusted to change the specification of the fragments (the potential design). One skilled in the art will recognize that other parameters may be incorporated into the automated design procedure as they are desired. In addition, in some embodiments, a potential design, even though "unsuccessful" according to the criteria may be optionally output (not shown) along with an indication of the relative strength of that particular design. The evaluation and design adjustment process continues until the design criteria are met (or the adjustment possibilities are exhausted).

As mentioned, in one embodiment, the APSDS evaluates (or scores) a potential synthesis design by employing "ends-checking rules" to test whether all combinations of possible interference reactions will succeed. The ends-checking rules account for the possibility of approximate complementarity of ends generating incorrect ligations, as well as other empirical rules that may cause incorrect ligations to occur in practice. In one embodiment, the ends-checking rules determine whether a pair of ends are: (1) exactly complementary using Watson-Crick traditional base pair interactions; (2) approximately complementary using base pairing rules empirically established (for example, "C/T" and "T/G" pairings); and/or (3) contain "GC" components within the overhang, that is, a GC or GG or CC or CG sequence within the overhang. In addition, partial matches (for example, 3 of 4 base pairs and 6 of 9 base pairs) of exactly or approximately complementary are also determined. Note also that different overhang lengths may empirically indicate the use of different ends-checking rules. In one embodiment, the APSDS is structured so that new ends-checking rules can be dynamically incorporated into the system, as new rules are empirically discovered or as rules need to be modified. Thus, the APSDS is designed to be extensible and dynamic as more empirical evidence for producing correct synthesis designs is discovered.

When used with convergent synthesis, the APSDS examines each potential ligation pair of fragments using the ends-checking rules and varies the specification of the fragments of the potential design until a pair of fragments is found that ligates only in the desired fashion. When used with other types of synthesis, the APSDS can be extended to search all possible combinations of ends that may arise depending upon the ordering of the ligation reactions and the number of fragments involved. For example, when used with solid phase synthesis, each fragment is successively ligated to a previous (longer) fragment that is attached to a solid support until the final fragment is produced. In this case, the left end of the fragment attached to the solid support need not be examined, but the left and right ends of the shorter fragments are examined. In another example, for other design reasons, more than two fragments may be ligated at the same time, such as in one solution where there is not an even number of fragments. In this case, the ends-checking rules are deployed against all possible combinations of the ends of the total number of fragments being examined.

FIGS. 6A 6K are example schematic diagrams of the fragments tested and adjusted at each level in an example APSDS used to generate a convergent synthesis design. In summary, convergent synthesis builds up a long polynucleotide sequence by successively ligating pairs of fragments, and then ligating, by pairs, the products of those reactions until there are no more pairs to ligate. In order to build a design for convergent synthesis that will ligate as desired, the techniques used by the APSDS successively check the ligatability of each pair of "potential" fragments, beginning with the longest (target) sequence, and working "backwards" to fragments of a length that approximates the default desired oligo length. (The technique uses a binary decomposition approach.) The APSDS modifies a pair of fragments (thereby adjusting the joint) when the ligatability index indicates that the potential fragments will not ligate according to designated criteria. One skilled in the art will recognize that whether one discusses the adjustments in terms of fragment definitions (attributes) or in terms of the joint itself does not affect the functionality of technique--a change in one will necessarily affect the other.

Specifically, as can be observed in FIG. 3, the example target sequence denoted as Level E is divided initially into 10 fragments (9 fragments of size desired oligo length and a 10.sup.th fragment of the remaining portion of the sequence). These initial "potential" fragments are then tested for proper ligatability and adjusted as needed according to a protocol (i.e., in an order) discussed relative to FIGS. 6A 6K. In FIGS. 6A and 6B, the APSDS examines the joint between the two potential fragments that currently constitute Level D. The APSDS examines the ends that may ligate in constructing this joint, evaluates the synthesis criteria, and if the evaluation indicates success, then the APSDS establishes that the next largest fragment (the "fragment" of Level E) will synthesize properly and thus includes Level D, and potential fragments 0 and 1 in the resultant synthesis design. For purposes of examining the next level, the joint between Level D fragments 0 and 1 is considered frozen.

Similarly, in FIGS. 6C and 6D, the APSDS examines the ends that contribute to constructing a joint between potential fragments 0 and 1 of Level C and the ends that contribute to constructing a joint between potential fragments 2 and 3 of Level C, according to the synthesis criteria. If the synthesis criteria indicate that one of these joints will not be successful, then in FIG. 6C the joint is adjusted. As mentioned, in one embodiment the joint adjustment consists of changing the overhang type of the joint or moving the joint position left or right (thus changing the fragment definitions). Once the synthesis criteria indicate that the joints will successfully ligate at this level, the APSDS establishes that these next largest fragments (the fragments of Level D) will synthesize properly and thus includes Level C, and potential fragments 0, 1, 2 and 3 in the resultant synthesis design. The design constructed thus far is shown in FIG. 6D. For purposes of examining the next level, the joints between fragments 0 and 1 and fragments 2 and 3 of Level C are considered frozen.

In a similar fashion, the joints between the potential fragments of the remaining levels are examined, evaluated, and adjusted until the synthesis criteria indicates that a potential fragment pair meets the synthesis criteria and should be included in the design. FIG. 6E shows the evaluations and adjustments made to create Level BX; and FIG. 6F shows the resulting intermediary design including the fragments of Levels E through BX. FIG. 6G shows the evaluations and adjustments made to create Level B; and FIG. 6H shows the resulting intermediary design including the fragments of Levels E through B