# US 2010/0169026 A1 · Algorithms for sequence determination

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2010/0169026 A1
- Document ID: US20100169026A1
- Status: Published application
- Kind: A1
- Application number: 12/592,284
- Filed: 2009-11-20
- Published: 2010-07-01
- Earliest claimed: 2008-11-20
- Assignees: Pacific Biosciences of California, Inc.
- Inventors: Jon Sorenson; Susan Tang; Patrick Marks; Chen-Shan Chin
- Web page: https://patentgrep.com/patent/US20100169026A1

## CPC

- G16B 30/10: Sequence alignment; Homology search
- G16B 30/00: ICT specially adapted for sequence analysis involving nucleotides or amino acids
- G16B 30/20: Sequence assembly
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY

## Abstract

The present invention is generally directed to powerful and flexible methods and systems for consensus sequence determination from replicate biomolecule sequence data. It is an object of the present invention to improve the accuracy of consensus biomolecule sequence determination from replicate sequence data by providing methods for assimilating replicate sequence into a final consensus sequence more accurately than any one-pass sequence analysis system.

## Claims

### Claim 1

A method of determining a consensus sequence, comprising:
a) providing a plurality of replicate sequence reads;
b) aligning the plurality of replicate sequence reads to generate a multiple sequence alignment;
c) determining a percent of majority calls for each position in a region of interest in the multiple sequence alignment;
d) based upon the results of step c, identifying at least one set of sequential positions in the region of interest, wherein all positions in the set have a percent of majority calls below a threshold; and
e) using information from the replicate sequence reads for the sequential positions in a statistical pattern classification algorithm to generate a consensus sequence spanning the sequential positions.

### Claim 2 (depends on 1)

The method of claim 1, wherein the consensus sequence is a polynucleotide sequence.

### Claim at position 3

3-5. (canceled)

### Claim 6 (depends on 1)

The method of claim 1, wherein the replicate sequence reads are provided by repeatedly sequencing a single molecule.

### Claim 7 (depends on 1)

The method of claim 1, wherein the replicate sequence reads are provided by simultaneously or sequentially sequencing multiple molecules, each of which comprises at least a portion of said region of interest.

### Claim at position 6

8-14. (canceled)

### Claim 15 (depends on 1)

The method of claim 1, wherein the percent of majority calls is determined for each position in the region of interest by dividing a count for a most frequent call at a given position by a total number of replicate sequence reads that comprise the given position.

### Claim at position 8

16-17. (canceled)

### Claim 18 (depends on 1)

The method of claim 1, further comprising generating a path representation of the MSA.

### Claim 19 (depends on 18)

The method of claim 18, wherein the path representation is a directed acyclic graph.

### Claim 20 (depends on 18)

The method of claim 18, wherein the path representation is subjected to pruning heuristics.

### Claim 21 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm finds a shortest path for calls across the sequential positions.

### Claim 22 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm applies a likelihood model.

### Claim 23 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm applies an error model.

### Claim 24 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm applies a probabilistic graph model.

### Claim 25 (depends on 24)

The method of claim 24, wherein the probabilistic graph model comprises an all path probabilistic alignment.

### Claim 26 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm applies both a likelihood model and an error model.

### Claim 27 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm applies a likelihood model, an error model, and a probabilistic graph model.

### Claim 28 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm generates the consensus sequence by maximizing a likelihood based upon the replicate sequence reads.

### Claim 29 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm uses a context dependent alignment model parameter based upon a whole genome multiple sequence alignment to generate the consensus sequence.

### Claim 30 (depends on 1)

The method of claim 1, wherein the statistical pattern classification algorithm applies a dynamic Bayesian network.

### Claim 31 (depends on 30)

The method of claim 30, wherein the dynamic Bayesian network is a Hidden Markov Model.

### Claim 32 (depends on 1)

The method of claim 1, further comprising catenating the consensus sequence generated in step e with majority calls from outside the sequential positions but in the region of interest.

### Claim 33 (depends on 1)

The method of claim 1, wherein multiple sets of sequential positions are identified in step d and multiple consensus sequences for the multiple sets of sequential positions are generated in step e.

### Claim 34 (depends on 33)

The method of claim 33, further comprising catenating the multiple consensus sequences with majority calls from outside the multiple sequential positions but in the region of interest to generate a further consensus sequence for the region of interest.

### Claim 35 (depends on 33)

The method of claim 33, wherein a first threshold is used to identify at least a first of the multiple sets of sequential positions and a second threshold is used to identify at least a second of the multiple sets of sequential positions.

### Claim 36 (depends on 1)

The method of claim 1, wherein a set of consensus sequences for a single set of sequential positions is generated in step e, further wherein the set of consensus sequences includes at least a most probable consensus sequence and a second most probable consensus sequence.

### Claim 37 (depends on 1)

The method of claim 1 iteratively applied, wherein in a first iteration only data of highest quality from the replicate sequence reads is used, and in a subsequent iteration additional data from the replicate sequence reads is used.

### Claim 38 (depends on 37)

The method of claim 37, wherein in a first iteration only the replicate sequence reads of highest quality are aligned in step b, and in a subsequent iteration additional of the replicate sequence reads are aligned in step b.

### Claim 39 (depends on 37)

The method of claim 37, wherein in a first iteration a first threshold is used to identify a first set of sequential positions, and in a subsequent iteration a second threshold is used to identify a second set of sequential positions, wherein the second threshold is more stringent than the first threshold.

### Claim 40 (depends on 2)

The method of claim 2, wherein the method is a computer-implemented method.

### Claim at position 32

41-43. (canceled)

### Claim 44

A method for determining a best call at a position in a sequence, comprising:
a) determining a set of experimental calls at a first position in a plurality of replicate sequences;
b) determining a set of experimental calls at one or more positions adjacent to the first position in the plurality of replicate sequences;
c) using the results from steps a and b in a statistical pattern classification algorithm to generate a best call for the first position.

### Claim 45

A method for determining a consensus sequence spanning a region of interest, comprising:
a) providing a data set comprising a plurality of replicate sequence reads comprising calls for each position in the region of interest;
b) identifying successive positions in the region of interest for which the data set contains at least one ambiguous call;
c) inputting the calls from the successive positions identified in step b into a statistical pattern classification algorithm to determine a best call at each of the successive positions; and
d) catenating the result of step c with calls from positions in the region of interest that are outside the successive positions identified in b, thereby determining a consensus sequence spanning the region of interest.

### Claim 46

(canceled)
