TWI430128B - Method for mining biclusters from dna microarray data using condition-enumeration tree - Google Patents

Description

Method for exploring two-way grouping in DNA microarray data by using conditional enumeration tree

The invention relates to a method for exploring two-way grouping in a data, in particular, a method for exploring two-way grouping in a DNA microarray data by using a conditional enumeration tree.

Sub-space grouping techniques for gene and experimental conditions in gene microarrays have been shown to help understand such functions as gene function, gene regulation, cellular processes, and cell subtypes. In general, in most cases, a disease is made up of multiple genes, so researchers are constantly trying to find out the similarities of certain genes under certain conditions as a basis for judging the disease.

In the prior art, in the method of supporting subspace grouping in gene microarrays, common methods such as pCluster and zCluster are subspace groupings for finding out that certain genes have consistent performance under certain conditions. However, both of these conventional methods involve very time consuming steps, namely constructing the largest set of dimensions of a gene pair and distributing the genetic information on each node of the prefix tree. Therefore, the conventional sub-space grouping method of the gene microarray of pCluster and zCluster has a complicated distribution process and must use a large amount of search space and search time. There is also a conventional MicroCluster method that converts the problem into a graph problem, but needs to solve an NP-Complete problem called Maximal Clique.

Therefore, it is necessary to provide an innovative and progressive method for enumerating two-dimensional clustering in DNA microarray data to solve the above problems.

The present invention provides a method for exploring two-way clustering in a DNA microarray data using a conditional enumeration tree, wherein the DNA microarray is an array of a plurality of genes and a plurality of conditional information of each gene, the genes having an order Gene number, the condition information has a sequential condition number, the method comprising the following steps: (a) calculating a plurality of conditional pair maximum dimension sets and a conditional count table, wherein each condition pair maximum dimension set includes two condition information and a plural number a bit string, the conditional count table includes condition information, condition information, the number of occurrences of the condition to the largest dimension set, and conditional information on the condition set of the maximum dimension set; (b) based on a minimum The condition information value is used to delete part of the condition information in the condition count table; and (c) the condition set table for the largest dimension set and the deletion according to the conditions, and a conditional enumeration is established by using a region breadth priority and a global depth priority method The tree lists the condition sets of the nodes of the tree and the set of related gene bit strings by calculating the condition.

The invention utilizes the conditional enumeration tree to explore the two-way grouping method in the DNA microarray data, does not need to construct the largest dimension set of the gene pair in the conventional technology, can avoid the complicated distribution process, and utilizes the concept of hash combination to efficiently Find out the results of two-way grouping. In addition, a conditional enumeration tree is established by using a region breadth priority and a global depth priority method. Thereby, the method of the invention for exploring two-way grouping by using the conditional enumeration tree in the DNA microarray data can avoid the complicated distribution process, and can greatly reduce the search space and the search time.

Moreover, the method for exploring two-way grouping in the DNA microarray data by using the conditional enumeration tree can be applied to a biological database analysis website set up on the World Wide Web (WWW), and the user can use the method of the invention to the gene microarray. Data for cluster analysis, or research units (eg, biomedical technology companies that study gene microarrays) can use clustering results to help understand the relationship between genes.

The method for exploring two-way clustering in the DNA microarray data using the conditional enumeration tree is described below with reference to the drawings. Figure 1 shows a schematic representation of DNA microarray data in accordance with an embodiment of the present invention. In an embodiment of the invention, the DNA microarray is an array of a plurality of genes and a plurality of conditional information for each gene, the genes having a sequence of gene numbers, the conditional information having a sequential condition number.

First, the method of the present invention calculates a plurality of condition-to-maximum dimension sets and a conditional count table, wherein each condition-to-maximum dimension set includes two conditional information and a plurality of bit-character strings, and the conditional count table includes condition information and condition information. The number and condition of the occurrence of the condition for the largest set of dimensions are related to the set of related conditions of the set of conditions to the largest set of dimensions.

The above steps of calculating the conditions for the largest dimension set, first calculating the difference of the condition information of each numbered gene under each two condition numbers, referring to FIG. 2, which shows the calculation of each number of genes in a and b conditions The difference between the condition information under the number. Then, the difference values are arranged from small to large as a difference, and the corresponding genes are arranged according to the difference. Referring to FIG. 3, the difference is arranged from small to large as a difference. Sort.

Then, according to a tolerance error value, the difference is sorted into groups, and the corresponding gene numbers are sorted according to the difference of the group, and the difference sequence of the group is represented by the bit string. Referring to FIG. 4, the genes showing the difference between the a and b condition information groups are gene numbers 0, 2, 4, and 6, so that the bits in the corresponding bit string are 0, 2, 4, 6 is 1, the other bits are 0; the genes in the group where the difference between the a and b condition information is 10 are the gene numbers 1, 3, and 5, so the bits 1 and 3 in the corresponding bit string are , 5 is 1, and the other bits are 0.

And determining, according to a minimum genetic value (NR), whether to delete the bit string to calculate the conditional pair of maximum dimensions, each condition pair maximum dimension set including a condition pair and a maximum dimension set, each condition The pair is the condition information of each corresponding difference, and each of the largest dimension sets is the bit string. In this embodiment, the minimum gene value (NR) is 3, and if the number of bits of the bit string is 1 and the number is less than 3, the bit string is deleted. The above two-bit string (1010101 and 0101010) is not less than 3 and is not deleted. Then the a and b condition pair maximum dimension set includes two conditional information a and b and a two-bit string (1010101 and 0101010).

The condition-to-maximum dimension set of all the two conditional information in the embodiment is calculated according to the above method. In this embodiment, the conditional count table is calculated for the largest dimension set according to the conditions, as shown in FIG. 5. Calculating, according to the conditions, the number of occurrences of each condition information on the condition set of the maximum dimension set for the maximum dimension set, and calculating the conditional alignment of each condition information to the maximum dimension set of the condition Another conditional information is its set of related conditions. The conditional count table includes condition information, condition information, the number of occurrences of the conditions for the largest set of dimensions, and conditional information on the set of related conditions of the condition to the largest set of dimensions. For example, the condition information a appears in the maximum dimension set of the conditions is 4, and the related condition set is b, c, d, e.

Then, part of the condition information is deleted in the condition count table according to a minimum condition information value (NC). If the condition information is less than 1 when the number of occurrences of the maximum dimension set is less than the minimum condition information value, the condition information and the condition information in the relevant condition set of other condition information are deleted. Referring to FIG. 6, which is a schematic diagram showing the deletion of partial condition information in the condition counter table, in the embodiment, the minimum condition information value (NC) is 3, because the condition information d appears in the maximum dimension set of the conditions. If the value of 1 is less than the minimum condition information minus 1 (NC-1), the condition information d and the condition information in the relevant condition set of other condition information are deleted, for example, the condition in the relevant condition set of the condition information a Information d.

Then, according to the conditions, for the maximum dimension set and the deleted condition count table, a conditional enumeration tree is established by using a local breadth-first within Global Depth-first method to calculate a condition. List the condition sets of nodes of the tree and the set of related gene bit strings. Referring to FIG. 7, the conditional enumeration tree is established by displaying the maximum dimension set and the deleted condition count table according to the conditions.

Referring to Figures 8 and 9, there is shown the steps of establishing a conditional enumeration tree and its hierarchy using a region breadth-first and global depth-first method. The following describes the steps for establishing the conditional enumeration tree. First, the condition information of the conditional count table is the first layer node of the conditional enumeration tree, as shown in step 1 of FIG. 8 and FIG. 9 (step 1). Then, one of the nodes is selected as a basic join node (join-base), the other nodes are placed in other join nodes, and the basic join node is paired with other nodes in the other join nodes to generate at least one extension node (expansion) The extended node has at least two condition information, and the extended node is disposed under the selected node. As shown in step 2 (step 2) of FIG. 8 and FIG. 9, for example, a is a basic join node (join-base), and other nodes (b, c, e) are placed in other join nodes (others), and the basic join is performed. The node is paired with other nodes in the other joining nodes to generate three extended nodes (ab, ac, ae), the extended node has two condition information, and the extended nodes (ab, ac, ae) are set at the selected node (a ) The next layer (the second layer, level 2).

One of the extended nodes is selected as a basic joining node, and the other extended nodes are placed in other joining nodes, and the basic joining node is paired with other extended nodes in the other joining nodes to generate at least one extended node, and the extended node has at least two condition information. And the extension node is disposed under the selection extension node. As shown in step 3 (step 3) of FIG. 8 and FIG. 9, for example, (ab) is selected as a basic joining node, and other extended nodes (ac, ae) are placed in other joining nodes, and the basic joining node and other joining nodes are basically added. The other extended nodes are paired to generate two extended nodes (abc, abe), each extended node has three conditional information, and the second extended node (abc, abe) is disposed under the selected extended node (ab) (the first Three levels, level 3).

Referring to FIG. 8 and FIG. 9, the above steps are repeated to establish each node of the conditional enumeration tree, and the condition information of each node is calculated as a condition set.

Then, the condition information in the condition set of the node of the tree is enumerated by using the condition, and the corresponding condition information is used to calculate the relevant genetic bit of the node of the conditional enumeration tree by using the condition information to the plurality of bit strings of the maximum dimension set. String collection. In this embodiment, the interaction logic and the (AND) operation (X) corresponding condition information may be used to calculate a plurality of bit strings of the maximum dimension set by the conditions, and the relevant genetic bits of the node of the conditional enumeration tree may be calculated. String collection.

With reference to the conditions of FIG. 5 for the maximum dimension set and FIG. 10, taking the node abc as an example, the set of related gene bit strings of the node abc can be expressed by the following equation: GS _abc = GS _ab X GS _ac X GS _bc , where GS _Ab is the bit string of the ab condition to the largest dimension set (1010101, 0101010), GS _ac is the ac condition to the maximum dimension set of the bit string (1010100, 0101010), GS _bc is the bc condition pair maximum The bit string of the set of dimensions (1010100, 0101010). The interactive logic AND (X) GS _ab , GS _ac and GS _bc are calculated as the correlation gene bit string set GS _abc of the node abc (1010100, 0101010). Similarly, the node abe can calculate its related gene bit string set GS _{abe by the} above method.

Since the method of the present invention establishes the conditional enumeration tree by using a region breadth-first and global depth-first method, another method can be used to calculate a set of related gene-bit strings of nodes of the conditional enumeration tree. The conditional enumeration tree has a condition set in the node of the tree having n condition information, and the first condition information is interactively ANDed to the n-1th condition information to the plurality of bits of the maximum dimension set of the condition The string and the first condition information to the n-2th condition information and the nth condition information in the plurality of bit strings of the condition to the largest dimension set, and the n-1th condition information and the nth The condition information is used to calculate a set of related gene bit strings of nodes of the conditional enumeration tree for the plurality of bit strings of the conditions of the maximum dimension set.

As shown in FIG. 11 , taking the node abcd as an example, the set of related gene bit strings of the node abcd can be expressed by the following formula: GS _abcd = GS _abc X GS _abd X GS _cd , because the method of the present invention utilizes regional breadth priority and global domain The depth-first method establishes the conditional enumeration tree, and the node abc and the node abd are established before the node abcd is established, so the above method can be used, and the interaction logic and the (AND) operation are fixed to the second, regardless of the condition information of the node. To reduce computing time and computational complexity.

Since the conventional interaction logic and (AND) operation requires interaction of all the bit strings in the set of operations, as shown in FIG. 12, there are four bit strings in the set A, and four bits in the set B. Meta-string, if set A and set B perform interactive logic and (AND) operations, 16 logical AND operations must be performed. If the similar bit strings in the set A are grouped, the similar bit strings in the set B are grouped, and the bit strings of the similar groups in the set are subjected to interactive logic and (AND) operations, as shown in FIG. To perform the interactive logic AND operation with the similar group of the set A and the similar group of the set B, only 8 logical AND operations are required.

Using the method of the similar group described above, the present invention can use another method to calculate a set of related gene bit strings of nodes of the conditional enumeration tree. And grouping a plurality of bit strings of the largest dimension set by the corresponding condition information to create a signature table having at least one signature bit string (Sig) and at least One bit string set (BSS), and the sign bit string of the signature table corresponding to the condition information is logically ANDed, if the number of logical AND (1) results is greater than one The minimum gene value is used to perform an interactive logic and (AND) operation on the bit string in the bit string set of the signature table corresponding to the condition information, and the relevant gene bit string of the node of the conditional enumeration tree is calculated. set.

Referring to FIG. 14, if an x node has three bit strings (1010101, 1010010, 0101010) in the maximum dimension set, the three bit strings are grouped to create a signature table. First, the first bit string (1010101) is first placed in the signature bit string (Sig) and the bit string set (BSS) of the first group. The second bit string (1010010) is logically ORed with the signature bit string (Sig, ie, the first bit string 1010101), and the logical OR (OR) operation result is calculated as one. The number, minus the number of 1 of the signature bit string (Sig), if less than or equal to a threshold T (for example, T is 1), it means that it can be the same group, and logical OR The operation result replaces the signature bit string (Sig), and the second bit string (1010010) is added to the bit string set (BSS).

Then, the third bit string (0101010) is logically ORed with the signature bit string (Sig, ie 1010111), and the number of logical OR (OR) results is calculated, and then subtracted. To sign the number of the first bit string (Sig), which is greater than the threshold, indicating that the third bit string is not the same as the first bit string and the first bit string. The same group, so the third bit string is the signature bit string (Sig) of another group, and is the bit string set (BSS) of another group.

Referring to FIG. 15, which shows the second signature table of the sub-group, when the operation of the two signature table is performed, the first signature bit string (1010101) and the second signature table of the first signature table S1. The first signature bit string of S2 (1010100) performs a logical AND operation. If the number of logical AND (AND) operations is greater than the minimum genetic value (NR), the group of the two is similar. Then, the bit string in the set of bit strings of the first signature table and the second signature table is subjected to an interactive logic and (AND) operation. By using the above-described grouping method, the number of logical AND operations can be reduced, and the operation time and operation complexity can be reduced.

When the conditional enumeration tree is established, the method of the present invention further includes a first limiting step for determining that if the maximum condition information number of all nodes under the node is less than the minimum condition information value, then the node is not required to be established. node. Referring to FIG. 16, the maximum condition information number of all nodes under the node c is 2, which is smaller than the minimum condition information value 3 of the embodiment. Therefore, the node ce under the node c does not need to be established.

The method of the present invention further includes a second limiting step, configured to determine that all condition information of a first determining node includes all condition information of a second determining node, and the related genetic bit string set of the first determining node Equal to the set of related gene bit strings of the second determining node, it is not necessary to establish a node under the second determining node. Referring to FIG. 17, if all the condition information (abce) of the first judgment node Y includes all condition information (ac) of the second judgment node X, and the related gene bit string set of the first judgment node Y (1010100) , 0101010) is equal to the set of related gene bit strings (1010100, 0101010) of the second judgment node X, and it is not necessary to establish the node ace under the second judgment node X.

The method of the present invention further includes a third limiting step for determining that all condition information of a first determining node includes all condition information of all nodes under a second determining node, although the relevant genetic bit of the first determining node The set of metastrings is not equal to the set of related gene bit strings of the second determining node, but if the set of related gene bit strings of the first determining node is equal to the set of related gene bit strings of a third determining node, And all the condition information of all the nodes under the second determining node includes all the condition information of the third determining node, and the node under the second determining node does not need to be established.

Referring to FIG. 18, if all condition information abce of the first judgment node Y includes all condition information ace of all nodes under the second judgment node X, although the first judgment node Y has a related gene bit string set (1010100, 0101010) ) is not equal to the set of related gene bit strings of the second judgment node X (1111111), but if the correlation gene bit string set of the first judgment node Y (1010100, 0101010) is equal to the third judgment node Z a set of gene bit strings (1010100, 0101010), and all condition information ace of all nodes under the second judgment node X includes all condition information ae of the third judgment node Z, and the second judgment node does not need to be established. Node ace under X.

The invention utilizes the conditional enumeration tree to explore the two-way grouping method in the DNA microarray data, does not need to construct the largest dimension set of the gene pair in the conventional technology, can avoid the complicated distribution process, and utilizes the concept of hash combination to efficiently Find out the results of two-way grouping. In addition, a conditional enumeration tree is established by using a region breadth priority and a global depth priority method. Thereby, the method of the invention for exploring two-way grouping by using the conditional enumeration tree in the DNA microarray data can avoid the complicated distribution process, and can greatly reduce the search space and the search time.

Moreover, the method for exploring two-way grouping in the DNA microarray data by using the conditional enumeration tree can be applied to a biological database analysis website set up on the World Wide Web (WWW), and the user can use the method of the invention to the gene microarray. Data for cluster analysis, or research units (eg, biomedical technology companies that study gene microarrays) can use clustering results to help understand the relationship between genes.

The above embodiments are merely illustrative of the principles and effects of the invention and are not intended to limit the invention. Therefore, those skilled in the art can make modifications and changes to the above embodiments without departing from the spirit of the invention. The scope of the invention should be as set forth in the appended claims.

1 shows a schematic diagram of DNA microarray data of an embodiment of the present invention;

Figure 2 is a diagram showing the difference between the condition information of each numbered gene under the condition number a and b;

FIG. 3 is a schematic diagram showing the ranking of the difference values from small to large as a difference order;

4 shows a schematic diagram of generating a corresponding bit string according to the difference ranking;

FIG. 5 is a schematic diagram showing the calculation of the conditional count table for the largest set of dimensions according to the conditions;

Figure 6 shows a schematic diagram of deleting part of the condition information in the condition count table;

FIG. 7 is a schematic diagram showing the establishment of the conditional enumeration tree according to the conditional count table after deletion;

8 and FIG. 9 are diagrams showing the steps of establishing a conditional enumeration tree and its hierarchy by using a region breadth-first and a global depth-first method;

FIG. 10 is a schematic diagram showing the calculation of a set of related gene bit strings by using an interactive logic and an AND operation;

FIG. 11 is a schematic diagram showing the calculation of a set of related gene bit strings by using a quadratic interaction logic and an AND operation;

Figure 12 shows a schematic diagram of conventionally calculating all bit strings in a set using interactive logic and (AND) operations;

Figure 13 is a diagram showing the calculation of all bit strings of similar groups in a set using interactive logic and (AND) operations;

Figure 14 is a schematic diagram showing the use of grouping to create a signature table for an x-node;

FIG. 15 is a schematic diagram showing operations performed by the second signature table of the group;

Figure 16 is a schematic view showing the first limiting step of the present invention;

Figure 17 is a view showing a second limiting step of the present invention;

Figure 18 shows a schematic diagram of a third limiting step of the present invention.

(no component symbol description)

Claims (12)

A method for exploring two-way clustering in a DNA microarray data using a conditional enumeration tree, wherein the DNA microarray is an array of a plurality of genes and a plurality of conditional information of each gene, the method comprising the steps of: (a) Calculating a plurality of condition-to-maximum dimension sets and a conditional count table, wherein each condition-to-maximum dimension set includes two condition information and at least one meta-word string, the condition count table including condition information, condition information, and the conditional maximum The number and condition information of the dimension set appearing on the condition set of the maximum dimension set; (b) deleting part of the condition information in the condition count table according to a minimum condition information value; and (c) according to the condition For the maximum dimension set and the deleted condition count table, a conditional enumeration tree is established by using a region breadth-first and global depth-first method to calculate a condition set of nodes of the conditional enumeration tree and a set of related gene bit strings. The method of claim 1, wherein calculating the condition-to-maximum dimension set in step (a) comprises the steps of: (a1) the genes have a sequence of gene numbers, and the condition information has a sequence of condition numbers, each of which is calculated The difference between the condition information of the numbered gene under each of the two condition numbers; (a2) the difference values are arranged from small to large as a difference order, and the corresponding genes are arranged according to the difference; (a3) According to a tolerance error value, the difference is sorted into groups, and the corresponding gene numbers are sorted according to the difference of the group, and the difference is sorted by the bit string; and (a4) is determined according to a minimum gene value. Whether to delete the bit string to calculate the condition pair to the largest dimension set, each condition pair maximum dimension set includes a condition pair and a maximum dimension set, and each condition pair is a condition of each corresponding difference Information, each largest dimension set is the bit string. The method of claim 2, wherein calculating the condition count table in step (a) comprises the step of: (a5) calculating a number of occurrences of each condition information in a condition pair of the condition for the largest set of dimensions; and A6) Calculate each condition information in the conditional pair of the conditional pair of the largest dimension set as its related condition set. The method of claim 1, wherein in the step (b), if the condition information is that the number of occurrences of the maximum dimension set is less than the minimum condition information value minus 1, the condition information and other condition information are deleted. The condition information in the relevant condition set. The method of claim 1, wherein the establishing the conditional enumeration tree in the step (c) comprises the steps of: (c1) counting the condition information of the table by the condition as a first layer node of the condition enumeration tree; (c2) One of the nodes is selected as a basic joining node, the other nodes are placed in other joining nodes, and the basic joining node is paired with other nodes in the other joining nodes to generate at least one extended node, the extended node has at least two condition information, and the extended node Set at a layer below the selection node; (c3) select one of the extension nodes as a basic join node, and the other extension nodes are placed in other join nodes, and the basic join node is paired with other extension nodes in other join nodes to generate at least one Extending the node, the extended node has at least two conditional information, and the extended node is disposed under the selected extended node; and (c4) repeating steps (c3) and (c2) to establish each node of the conditional enumeration tree, and The condition information of each node is calculated as a condition set. The method of claim 1, wherein in the step (c), the condition information in the condition set of the node of the tree is enumerated by using the condition, and the corresponding condition information is used in the plurality of bit strings of the maximum dimension set in the condition And calculating a set of related gene bit strings of nodes of the conditional enumeration tree. The method of claim 6, wherein in step (c), the conditional enumeration tree is calculated by using an interactive logic and (AND) operation corresponding condition information on the plurality of bit strings of the maximum dimension set of the conditions. A collection of related gene bit strings for a node. The method of claim 6, wherein in the step (c), the condition set of the node of the conditional enumeration tree has n condition information, and the first condition information is interactively ANDed to the n-1th The condition information is a plurality of bit strings and the first condition information of the condition to the largest dimension set to the n-2th condition information and the nth condition information is a plurality of bits of the condition to the largest dimension set The string, and the n-1th condition information and the nth condition information are used in the plurality of bit strings of the condition to the largest dimension set, and the set of related gene bit strings of the node of the conditional enumeration tree is calculated. The method of claim 6, wherein in step (c), the corresponding condition information is grouped into the plurality of bit strings of the largest dimension set by the conditions, and a signature table is created, the signature table having at least one The signature bit string and the at least one meta string set, and the signature bit string of the signature table corresponding to the condition information is logically ANDed, if the logical AND result is 1 If the number is greater than a minimum gene value, then the bit string in the set of bit strings corresponding to the signature table of the condition information is subjected to an interactive logic and (AND) operation, and the node of the conditional enumeration tree is calculated. A collection of gene bit strings. The method of claim 5, wherein in step (c), a first limiting step is further included to determine that if the maximum condition information of all nodes under the node is less than the minimum condition information value, the The node under the node. The method of claim 5, wherein in step (c), a second limiting step is further included to determine that all condition information of a first determining node includes all condition information of a second determining node, and the If the set of related gene bit strings of the determining node is equal to the set of related gene bit strings of the second determining node, then the node under the second determining node does not need to be established. The method of claim 5, wherein in step (c), a third limiting step is further included to determine that all condition information of a first determining node includes all condition information of all nodes under a second determining node. The set of related gene bit strings of the first determining node is not equal to the set of related gene bit strings of the second determining node, but if the set of related gene bit strings of the first determining node is equal to a third Determining a set of related gene bit strings of the node, and all condition information of all nodes under the second determining node includes all condition information of the third determining node, and the node under the second determining node is not required to be established.