OperonPredictionFeatures

Full name SMRUCC.genomics.Model.OperonMapper.ContextModel.OperonPredictionFeatures Assembly SMRUCC.genomics.Model.OperonMapper Members 14

提取并计算操纵子预测所需的各种特征得分

To evaluate the contribution of selected features in operon prediction, we have calculated the numerical values of the features, And then used these values individually And in combination to train a classifier. The features used in our study are

(i) the intergenic distance,
(ii) the conserved gene neighborhood,
(iii) distances between adjacent genes' phylogenetic profiles,
(iv) the ratio between the lengths of two adjacent genes And
(v) frequencies Of specific DNA motifs in the intergenic regions.

00 Remarks

Operon prediction using both genome-specific and general genomic information

https://academic.oup.com/nar/article/35/1/288/2401876

01 Syntax

SMRUCC.genomics.Model.OperonMapper.ContextModel.OperonPredictionFeatures

02 Methods

Name	Overloads	Summary
CalculateAllFeatures	1	计算所有特征的综合函数包含前置校验(同链验证)及论文中所列全部6类特征的提取
CalculateKEGGSimilarity	1	基于KEGG通路和代谢反应连接性的功能相似性打分替代原有的GO功能相似性
CalculateIntergenicDistance	1	1. 计算基因间距离论文公式: DI = downstream_gene_start - (upstream_gene_end + 1) 并根据论文 Materials and Methods 应用 [-50, 250] 的截断值
CalculateNeighborhoodConservation	1	where L(gi, gj, Gk) is the loglikelihood of a gene pair to be neighbors in the kth genome Gk.
CalculatePhylogeneticDistanceHamming	1	For the Hamming distance between two genes A And B, we sum the number Of times that only A Or B Is found in the genome, DH=Sum([1,n], di)‚ where n Is the number of genomes, di=0…
CalculatePhylogeneticDistanceShannon	1	3. 计算系统发育距离 - 香农熵距离 (补充实现) 论文公式: DE = n - (n - DH) * Sqrt(E(p) / p) 其中 p 是0身份(两基因均不存在)的比例，E(p) = -plog(p) - (1-p)log(1-p)
CalculateLengthRatio	1	The score is calculated as the natural log of the length ratio of upstream gene And downstream gene, Or L = ln(li/lj), j = i + 1, whereas li And lj are the length of the genes.
CalculateMotifFrequency	1	5. 计算DNA基序频率论文公式: Fm = X / ((L-d+1) * p) X为观测次数，L为序列长度，d为基序长度，p为基于核苷酸频率计算的期望频率
CalculateGOSimilarity	1	6. 计算GO功能相似性 (修正为严格符合论文的路径最大匹配) 论文定义: SGO(gi,gj) = max s(Vi,Vj) s(Vi,Vj) 是由两个基因的GO术语分别诱导出的两条路径之间的共同术语数量取所有路径对中的最大值
AddAncestorTerms	1	递归添加祖先GO术语到集合中
GetDistanceGroup	1	判断相邻基因对属于哪个距离分组，用于后续选择特定的分类器模型论文 Results: 划分为 U40, U200, O200 进行子组训练

03 Fields

Name	Overloads	Summary
referenceGenomes	1	参考基因组列表
goHierarchy	1	GO层次结构字典：GO术语 -> 父术语列表
phylumProbabilities	1	基因存在概率字典

04 Members

method CalculateAllFeatures #

CalculateAllFeatures(GeneInfo, GeneInfo, String, Dictionary(Of Char, Double))

计算所有特征的综合函数包含前置校验(同链验证)及论文中所列全部6类特征的提取

Parameters

Name	Type	Description
`upstreamGene`	GeneInfo	上游基因信息
`downstreamGene`	GeneInfo	下游基因信息
`intergenicSequence`	`String`	基因间序列 (下游基因上游100nt)
`nucleotideFrequencies`	`Dictionary(Of Char, Double)`	核苷酸频率字典 (A, T, C, G -> 频率)

Returns

包含所有得分的 FeatureScores 对象；若基因不在同一链上则返回 Nothing

method CalculateKEGGSimilarity #

CalculateKEGGSimilarity(GeneInfo, GeneInfo, Dictionary(Of String, String()), MetabolicReaction())

基于KEGG通路和代谢反应连接性的功能相似性打分替代原有的GO功能相似性

Parameters

Name	Type	Description
`gene1`	GeneInfo	上游基因
`gene2`	GeneInfo	下游基因
`ecMaps`	`Dictionary(Of String, String())`	EC编号到KEGG Pathway的映射字典 [ec_number => map_id array]
`reactions`	`MetabolicReaction()`	全局代谢反应集合

Returns

综合功能相似性得分

method CalculateIntergenicDistance #

CalculateIntergenicDistance(GeneInfo, GeneInfo)

计算基因间距离

论文公式: DI = downstream_gene_start - (upstream_gene_end + 1) 并根据论文 Materials and Methods 应用 [-50, 250] 的截断值

Parameters

Name	Type	Description
`upstreamGene`	GeneInfo	上游基因信息
`downstreamGene`	GeneInfo	下游基因信息

Returns

经截断处理后的基因间距离

method CalculateNeighborhoodConservation #

CalculateNeighborhoodConservation(GeneInfo, GeneInfo)

where L(gi, gj, Gk) is the loglikelihood of a gene pair to be neighbors in the kth genome Gk. The log-likelihood score Is computed as the probability that gi And gj are neighbors within a distance dk(i,j) In Gk, Or L(gi, gj, Gk) = log Pij; Pij Is defined as follows:

(i) Pij = (1-pik)(1-pjk), if both genes are absent from genome Gk, (ii) Pij = (1-pik) pjk, if only gene j Is present In genome Gk, (iii) Pij = pik (1-pjk), if only gene i Is present In genome Gk, (iv) Pij = (pikpjkdk(i, j) (2Nk-dk(i,j)-1))/(Nk(Nk-1)), If genes i And j are present In genome Gk.

dk(ij) Is the number of genes between gi And gj; Nk Is the number Of genes In genome Gk; And pik Is the probability that gene gi Is present In genome Gk.

Remarks

计算基因邻域保守性 (Neighborhood Conservation)

Parameters

Name	Type	Description
`gene1`	GeneInfo	基因1信息
`gene2`	GeneInfo	基因2信息

Returns

邻域保守性总得分 S

method CalculatePhylogeneticDistanceHamming #

CalculatePhylogeneticDistanceHamming(GeneInfo, GeneInfo)

For the Hamming distance between two genes A And B, we sum the number Of times that only A Or B Is found in the genome, DH=Sum([1,n], di)‚ where n Is the number of genomes, di=0 if the orthologs of A And B are both present Or both absent in genome i, And di = 1 otherwise.

Remarks

计算系统发育距离 (Hamming Distance)

Parameters

Name	Type	Description
`gene1`	GeneInfo	基因1信息
`gene2`	GeneInfo	基因2信息

Returns

汉明距离得分

method CalculatePhylogeneticDistanceShannon #

CalculatePhylogeneticDistanceShannon(GeneInfo, GeneInfo, Int32)

计算系统发育距离 - 香农熵距离 (补充实现)

论文公式: DE = n - (n - DH) Sqrt(E(p) / p) 其中 p 是0身份(两基因均不存在)的比例，E(p) = -plog(p) - (1-p)*log(1-p)

Parameters

Name	Type	Description
`gene1`	GeneInfo	基因1信息
`gene2`	GeneInfo	基因2信息
`DH`	`Int32`	distance result of OperonPredictionFeatures.CalculatePhylogeneticDistanceHamming()

Returns

香农熵距离得分

method CalculateLengthRatio #

CalculateLengthRatio(GeneInfo, GeneInfo)

The score is calculated as the natural log of the length ratio of upstream gene And downstream gene, Or L = ln(li/lj), j = i + 1, whereas li And lj are the length of the genes.

Remarks

计算基因长度比 (Length Ratio)

Parameters

Name	Type	Description
`upstreamGene`	GeneInfo	-
`downstreamGene`	GeneInfo	-

method CalculateMotifFrequency #

CalculateMotifFrequency(String, String, Dictionary(Of Char, Double))

计算DNA基序频率

论文公式: Fm = X / ((L-d+1) * p) X为观测次数，L为序列长度，d为基序长度，p为基于核苷酸频率计算的期望频率

Parameters

Name	Type	Description
`intergenicSequence`	`String`	下游基因上游100nt的基因间序列
`motif`	`String`	待检测的DNA基序字符串
`nucleotideFrequencies`	`Dictionary(Of Char, Double)`	核苷酸频率字典 (A, T, C, G -> 频率)

Returns

基序的归一化频率得分

method CalculateGOSimilarity #

CalculateGOSimilarity(GeneInfo, GeneInfo)

计算GO功能相似性 (修正为严格符合论文的路径最大匹配)

论文定义: SGO(gi,gj) = max s(Vi,Vj) s(Vi,Vj) 是由两个基因的GO术语分别诱导出的两条路径之间的共同术语数量取所有路径对中的最大值

Parameters

Name	Type	Description
`gene1`	GeneInfo	基因1信息
`gene2`	GeneInfo	基因2信息

Returns

GO功能相似性最大得分

method AddAncestorTerms #

AddAncestorTerms(String, HashSet(Of String))

递归添加祖先GO术语到集合中

Parameters

Name	Type	Description
`term`	`String`	当前GO术语
`termSet`	`HashSet(Of String)`	用于收集祖先术语的HashSet

method GetDistanceGroup #

GetDistanceGroup(GeneInfo, GeneInfo)

判断相邻基因对属于哪个距离分组，用于后续选择特定的分类器模型论文 Results: 划分为 U40, U200, O200 进行子组训练

Parameters

Name	Type	Description
`upstreamGene`	GeneInfo	上游基因信息
`downstreamGene`	GeneInfo	下游基因信息

Returns

距离分组枚举值；如果两基因不在同一链上，则返回 Nothing

field referenceGenomes #

referenceGenomes

参考基因组列表

field goHierarchy #

goHierarchy

GO层次结构字典：GO术语 -> 父术语列表

field phylumProbabilities #

phylumProbabilities

基因存在概率字典

Remarks

基因存在概率字典：基因ID -> (门 -> 概率pik)