viral_seq 1.3.0 → 1.6.1

data/lib/viral_seq/seq_hash.rb

@@ -208,6 +208,31 @@ module ViralSeq
       return ViralSeq::SeqHash.new(sampled_nt, sampled_aa, sampled_qc, sampled_title, self.file)
     end
 
+    # return a new SeqHash object with given a range on the nt sequence position
+    # @param range [Range] range of positions on the nt sequence
+    # @return [ViralSeq::SeqHash] a sub SeqHash object
+
+    def nt_range(range)
+      dna_hash = self.dna_hash
+      new_hash = {}
+      dna_hash.each do |k,v|
+        new_hash[k] = v[range]
+      end
+      ViralSeq::SeqHash.new(new_hash)
+    end # end of #nt_range
+
+    # check the size range of the DNA sequences of the SeqHash object
+    # @return [Hash] Hash of {max: MAX_SIZE, min: MIN_SIZE}
+    
+    def check_nt_size
+      dna_hash = self.dna_hash
+      size_array = []
+      dna_hash.values.each do |v|
+        size_array << v.size
+      end
+      return { max: size_array.max, min: size_array.min }
+    end
+
     # write the nt sequences to a FASTA format file
     # @param file [String] path to the FASTA output file
     # @return [NilClass]
@@ -592,6 +617,98 @@ module ViralSeq
 
     alias_method :pm, :poisson_minority_cutoff
 
+    # calculate false detection rate for minority mutations
+    # Credit: Prof. Michael G. Hudgens from UNC-CH for providing the method for fdr calculation
+    # @param error_rate [Float] estimated sequencing error rate
+    # @return [Hash] pair of mutation frequency to false detection rate. (freq => fdr)
+    # @example calculate FDR for mutations that appeared twice in the sample dataset
+    #   my_seqhash = ViralSeq::SeqHash.fa('spec/sample_files/sample_sequence_for_poisson.fasta')
+    #   fdr_hash = my_seqhash.fdr
+    #   fdr_hash[2].round(5)
+    #   => 0.00726 # means that mutations appear twice have 0.007261748 chance to be caused by residual errors.
+
+    def fdr(error_rate = 0.0001)
+      sequences = self.dna_hash.values
+      if sequences.size == 0
+        return {}
+      else
+        seq_count = self.size
+        observed_hash = variant_for_poisson(sequences)
+        p_unadjusted = []
+        observed_hash.each do |k, v|
+          p_value = 1 - `Rscript -e "cat(pbinom(#{k}-1, #{seq_count}, #{error_rate}))"`.to_f # compute unadjusted exact p-value, ie under null, probability of observing observed_hash[k] or more extreme
+          p_unadjusted += Array.new(v, p_value)
+        end
+        p_fdr = `Rscript -e "cat(p.adjust(c(#{p_unadjusted.join(',')}), 'fdr'))"`.split("\s").count_freq.to_a # controls fdr. aka Benjamini-Hochberg correction
+        vars_pair = observed_hash.to_a
+        fdr_hash = Hash.new(0)
+        (0..(p_fdr.size - 1)).each do |i|
+          fdr_hash[vars_pair[i][0]] = p_fdr[i][0].to_f
+        end
+        return fdr_hash
+      end
+    end #end of #fdr
+
+    # analysis for the nt sequence variants.
+    # @return [Hash] An Hash with information of variant analysis. Key/values of the return object see /docs/variants_structure.pdf
+
+    def nt_variants
+      return_obj = {}
+      nt_hash = self.dna_hash
+      tcs_number = self.size
+      dl = ViralSeq::TcsCore.detection_limit(tcs_number)
+      fdr_hash = self.fdr
+      pm_cut_off = self.pm
+      con = self.consensus
+      return_obj[:tcs_number] = tcs_number
+      return_obj[:lower_detection_limit] = dl
+      return_obj[:pm_cut_off] = pm_cut_off
+      return_obj[:positions] = []
+      cis = {}
+
+      (0..(con.size - 1)).each do |p|
+        position_obj = {}
+        position_obj[:position] = p + 1
+        position_obj[:tcs_number] = tcs_number
+        position_obj[:lower_detection_limit] = dl
+        position_obj[:pm_cut_off] = (pm_cut_off == Float::INFINITY ? pm_cut_off.to_s : pm_cut_off)
+
+        nts = []
+        dna_hash.each do |n,s|
+          nts << s[p]
+        end
+        freq_hash = nts.count_freq
+        [:A, :C, :G, :T, :-].each do |k|
+          v = freq_hash[k.to_s]
+          position_obj[k] = {}
+          position_obj[k][:count] = v
+          if v > 0
+            if cis[[v, tcs_number]]
+              ci = cis[[v, tcs_number]]
+            else
+              ci = ViralSeq::Math::BinomCI.new(v, tcs_number)
+              cis[[v, tcs_number]] = ci
+            end
+            position_obj[k][:freq] = ci.mean.round(4)
+            position_obj[k][:freq_ci_low] = ci.lower.round(4)
+            position_obj[k][:freq_ci_high] = ci.upper.round(4)
+            position_obj[k][:greater_than_pm] = (v >= pm_cut_off ? true : false)
+            position_obj[k][:fdr] = fdr_hash[v]
+          else
+            position_obj[k][:freq] = 0
+            position_obj[k][:freq_ci_low] = 0
+            position_obj[k][:freq_ci_high] = 0
+            position_obj[k][:greater_than_pm] = false
+            position_obj[k][:fdr] = nil
+          end
+        end
+
+        return_obj[:positions] << position_obj
+      end
+
+      return_obj
+    end # end of nt_variants
+
 
     # align the @dna_hash sequences, return a new ViralSeq::SeqHash object with aligned @dna_hash using MUSCLE
     # @param path_to_muscle [String], path to MUSCLE excutable. if not provided (as default), it will use RubyGem::MuscleBio
@@ -1183,6 +1300,28 @@ module ViralSeq
       return new_sh
     end
 
+    # QC for each nucleotide sequence comparing with sample consensus for indels
+    # @return [Hash] object containing two SeqHash {no_indel: seq_hash, has_indel: seq_hash}
+
+    def qc_indel
+      con = self.consensus
+      dna_hash = self.dna_hash
+      names_passed = []
+      names_indel = []
+      dna_hash.uniq_hash.each do |seq, names|
+        if seq.compare_with(con) < 4
+          names_passed += names
+        elsif ViralSeq::Muscle.align(con, seq)[0]["-"]
+          names_indel += names
+        else
+          names_passed += names
+        end
+      end
+      return {no_indel: self.sub(names_passed),
+        has_indel: self.sub(names_indel)}
+    end # end of qc_indel
+
+
     # trim dna sequences based on the provided reference coordinates.
     # @param start_nt [Integer,Range,Array] start nt position(s) on the refernce genome, can be single number (Integer) or a range of Integers (Range), or an Array
     # @param end_nt [Integer,Range,Array] end nt position(s) on the refernce genome,can be single number (Integer) or a range of Integers (Range), or an Array

data/lib/viral_seq/tcs_core.rb

@@ -6,6 +6,10 @@ module ViralSeq
     class << self
 
       # methods to calculate TCS consensus cut-off based on the maximum numbers of PIDs and platform error rate.
+      # @see https://www.ncbi.nlm.nih.gov/pubmed/26041299 reference at Zhou et al. JVI 2016.
+      # @param m [Integer] PID abundance
+      # @param error_rate [Float] estimated platform error rate.
+      # @return [Integer] an abundance cut-off (Integer) for offspring Primer IDs.
 
       def calculate_cut_off(m, error_rate = 0.02)
         n = 0
@@ -280,6 +284,23 @@ module ViralSeq
         abort infor.red.bold
       end
 
+      # lower detection sensitivity for minority mutations given the number of TCS, calculated based on binomial distribution.
+      # R required.
+      # @param tcs_number [Integer] number of TCS
+      # @return [Float] lower detection limit
+      # @example calculate lower detection limit
+      #   ViralSeq::TcsCore.detection_limit(100)
+      #   => 0.0362
+
+      def detection_limit(tcs_number)
+        if ViralSeq::DETECT_SEN[tcs_number]
+          return ViralSeq::DETECT_SEN[tcs_number]
+        else
+          dl =  `Rscript -e "library(dplyr); ifelse(#{tcs_number} > 2, (binom.test(0,#{tcs_number})['conf.int'] %>% unlist %>% unname)[2] %>% round(4) %>% cat, 0)" 2>/dev/null`
+          dl.to_f
+        end
+      end
+
       private
 
       def unzip_r(indir, f)

data/lib/viral_seq/version.rb

@@ -2,6 +2,6 @@
 # version info and histroy
 
 module ViralSeq
-  VERSION = "1.3.0"
-  TCS_VERSION = "2.3.8"
+  VERSION = "1.6.1"
+  TCS_VERSION = "2.5.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: viral_seq
 version: !ruby/object:Gem::Version
-  version: 1.3.0
+  version: 1.6.1
 platform: ruby
 authors:
 - Shuntai Zhou
@@ -9,7 +9,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2021-08-30 00:00:00.000000000 Z
+date: 2022-02-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -172,6 +172,7 @@ files:
 - docs/dr.json
 - docs/sample_miseq_data/hivdr_control/r1.fastq.gz
 - docs/sample_miseq_data/hivdr_control/r2.fastq.gz
+- docs/variants_structure.pdf
 - lib/viral_seq.rb
 - lib/viral_seq/constant.rb
 - lib/viral_seq/enumerable.rb