done flexible parameters

tuncK · May 26, 2017 · 9843c76 · 9843c76
1 parent 1bfe9a6
commit 9843c76
Show file tree

Hide file tree

Showing 11 changed files with 200 additions and 103 deletions.
diff --git a/Db/Mouse.parameters.xml b/Db/Mouse.parameters.xml
@@ -117,4 +117,40 @@
 	<grr>CCGCAACATCCAGCATCGTG</grr>
   </onePCR>
 
+  <!-- otherSteps parameters -->
+  <!-- blastOtherSteps.m -->
+  <otherSteps>
+    <key1>ENS\w*T\d*</key1>
+	<thres>22</thres>
+	<querySize>50</querySize>
+    <blastArgs1>-S 2</blastArgs1>
+	<blastArgs2>-S 3</blastArgs2>
+	<grr>CCGCAACATCCAGCATCGTG</grr>
+	<T7r>CCCTATAGTGAGTCGTATTA</T7r>
+	<rRr>AGAGTGAGTAGTAGTGGAGT</rRr>
+	<rGr>GATGATGTAGTAGTAAGGGT</rGr>
+    <rBr>TGTGATGGAAGTTAGAGGGT</rBr>
+	<rIRr>GGAGTAGTTGGTTGTTAGGA</rIRr>
+  </otherSteps>
+
+  <!-- arna parameters -->
+  <!-- blastAbundantRNA.m -->
+  <arna>
+    <key1>ENS\w*T\d*</key1>
+	<seqNum>1000</seqNum>
+	<thres>30</thres>
+	<querySize>73</querySize>
+	<DbSize>200000</DbSize>
+    <blastArgs>-S 2</blastArgs>
+	<parallel>0</parallel>
+  </arna>
+
+  <!-- probeList parameters -->
+  <!-- generateProbeList.m -->
+  <probeList>
+    <key1>ENS\w*T\d*</key1>
+	<number>48</number>
+	<dir1>C:\FISHerMan\Db\Mouse.STList.fas</dir1>
+  </probeList>
+
 </parameters>
diff --git a/blastAbundantRNA.m b/blastAbundantRNA.m
@@ -3,15 +3,15 @@
 % homology of 15 nt or more.
 
 function [Header,Sequence,nonSequence,nonSequence2]...
-    =blastAbundantRNA(adapterList,Header,Sequence,nonSequence,nonSequence2,varargin)
+    =blastAbundantRNA(adapterList,Header,Sequence,nonSequence,nonSequence2,params)
 
-if length(varargin) >= 1
-    params = varargin{1};
-else
-    params = struct('species','Mouse','verbose',1,...
-        'thres',30,'querySize',73,'DbSize',2*10^5,'seqNum',1000,...
-        'blastArgs','-S 2','parallel', 0);
-end
+% if length(varargin) >= 1
+%     params = varargin{1};
+% else
+%     params = struct('species','Mouse','verbose',1,...
+%         'seqNum',1000,'thres',30,'querySize',73,'DbSize',200000,...
+%         'blastArgs','-S 2','parallel', 0);
+% end
 
 if isempty(nonSequence)
     nonSequence = Sequence;
@@ -21,28 +21,29 @@
     nonSequence2 = Sequence;
 end
 
-if params.verbose
+if params(1).verbose
     disp('removing non-specific probes against the abundant rna database');
 end
 
 %% Split one giant fasta file into smaller ones, so that parallel computing is possible
-if params.verbose
+if params(1).verbose
     disp('  spliting fasta files for parallel computing');
 end
 
-filePathList = blastFileSplit(Header, Sequence, params.seqNum);
+filePathList = blastFileSplit(Header, Sequence, params(1).seqNum, params);
 fileNum = length(filePathList);
 
 %% Blast mouse oligos against abundant rna
-eValue = bitScore2eValue(params.thres, params.querySize, params.DbSize);
+DbPath = [params(1).species '.abundantrnaDb.fas'];
+params(1).DbSize = getDbSize(DbPath);
 
-DbPath = [params.species '.abundantrnaDb.fas'];
-blastArgs = [params.blastArgs ' -e ' num2str(eValue)];
+eValue = bitScore2eValue(params(1).thres, params(1).querySize, params(1).DbSize);
+blastArgs = [params(1).blastArgs ' -e ' num2str(eValue)];
 
 blastData = {};
-if params.parallel
+if params(1).parallel
     poolobj = parpool;
-    verbose = params.verbose;
+    verbose = params(1).verbose;
     parfor k = 1:fileNum
         if verbose
             disp(['  blasting temporary file no. ' num2str(k)]);
@@ -57,12 +58,12 @@
     delete(poolobj);
 else
     for k = 1:fileNum
-        if params.verbose
+        if params(1).verbose
             disp(['  blasting temporary file no. ' num2str(k)]);
             startTime = tic;
         end
         blastData{k,1} = blastOp(filePathList{k}, DbPath, blastArgs);
-        if params.verbose
+        if params(1).verbose
             totalTime = toc(startTime);
             disp(['  elapsed time is ' num2str(totalTime) ' seconds']);
         end
@@ -95,8 +96,8 @@
 nonSequence2(seqDelete)= [];
 
 %% Check how many transcripts are left after this step of screening
-[geneNumLeft,geneNumDelete]=checkTranscriptsLeft(adapterList,Header);
-if params.verbose
+[geneNumLeft,geneNumDelete]=checkTranscriptsLeft(adapterList,Header,params);
+if params(1).verbose
     disp([num2str(geneNumDelete) ' out of ' num2str(geneNumLeft+geneNumDelete)...
         ' FISH escaped FISHerMan''s net']);
 end

diff --git a/blastOtherSteps.m b/blastOtherSteps.m
@@ -1,17 +1,18 @@
 function [probeHeader,probeSequence,probeSequence3Seg,probeSequenceCore]...
-    =blastOtherSteps(adapterList,probeHeader,probeSequence,probeSequence3Seg,probeSequenceCore,varargin)
-
-if length(varargin) >= 1
-    params = varargin{1};
-else
-    params = struct('species','Mouse','verbose',1,'keys','ENS\w*T\d*',...
-        'thres',22,'querySize',50,'blastArgs','',...
-        'grr','CCGCAACATCCAGCATCGTG','T7r','CCCTATAGTGAGTCGTATTA',...
-        'rRr','AGAGTGAGTAGTAGTGGAGT','rGr','GATGATGTAGTAGTAAGGGT',...
-        'rBr','TGTGATGGAAGTTAGAGGGT','rIRr','GGAGTAGTTGGTTGTTAGGA');
-end
-
-if params.verbose
+    =blastOtherSteps(adapterList,probeHeader,probeSequence,probeSequence3Seg,probeSequenceCore,params)
+
+% if length(varargin) >= 1
+%     params = varargin{1};
+% else
+%     params = struct('species','Mouse','verbose',1,'keys','ENS\w*T\d*',...
+%         'thres',22,'querySize',50,...
+%         'blastArgs1','-S 2','blastArgs2','-S 3',...
+%         'grr','CCGCAACATCCAGCATCGTG','T7r','CCCTATAGTGAGTCGTATTA',...
+%         'rRr','AGAGTGAGTAGTAGTGGAGT','rGr','GATGATGTAGTAGTAAGGGT',...
+%         'rBr','TGTGATGGAAGTTAGAGGGT','rIRr','GGAGTAGTTGGTTGTTAGGA');
+% end
+
+if params(1).verbose
     disp('removing probes that non-specifically bind to 2nd PCR primers and other probes');
 end
 
@@ -21,17 +22,17 @@
 
 simpleHeader = probeHeader;
 for n = 1:length(probeHeader)
-    pos = regexp(probeHeader{n,1}, params.keys, 'end');
+    pos = regexp(probeHeader{n,1}, params(1).keys, 'end');
     simpleHeader{n,1} = probeHeader{n,1}(1:pos);
 end
 
 seqDelete = [];
 for n = 1:length(adapterHeader)
-    if params.verbose% && mod(n, 1000) == 1
+    if params(1).verbose% && mod(n, 1000) == 1
         disp(['  working on trancript ' adapterHeader{n,1}]);
     end
     temp...
-        =blastOneTranscript(adapterHeader{n,1},adapterSequence{n,1},simpleHeader,probeSequenceCore);
+        =blastOneTranscript(adapterHeader{n,1},adapterSequence{n,1},simpleHeader,probeSequenceCore,params);
     seqDelete = [seqDelete temp];
 end
 
@@ -41,8 +42,8 @@
 probeSequenceCore(seqDelete)= [];
 
 %% Check how many transcripts are left after this step of screening
-[geneNumLeft,geneNumDelete]=checkTranscriptsLeft(adapterList,probeHeader);
-if params.verbose
+[geneNumLeft,geneNumDelete]=checkTranscriptsLeft(adapterList,probeHeader,params);
+if params(1).verbose
     disp([num2str(geneNumDelete) ' out of ' num2str(geneNumLeft+geneNumDelete)...
         ' FISH escaped FISHerMan''s net']);
 end

diff --git a/generateProbeList.m b/generateProbeList.m
@@ -1,19 +1,19 @@
-function probeList=generateProbeList(adapterList,probeHeader,probeSequence,varargin)
+function probeList=generateProbeList(adapterList,probeHeader,probeSequence,params)
 
-if length(varargin) >= 1
-    params = varargin{1};
-else
-    params = struct('species','Mouse','verbose',1,'number',48,...
-        'keys','ENS\w*T\d*');
-end
+% if length(varargin) >= 1
+%     params = varargin{1};
+% else
+%     params = struct('species','Mouse','verbose',1,'keys','ENS\w*T\d*',...
+%         'number',48,'specialTranscripts','C:\FISHerMan\Db\Mouse.STList.fas');
+% end
 
 %% Remove transcripts without enough probes
-if params.verbose
+if params(1).verbose
     disp('generating the list of probes to order');
     disp('  removing transcripts without enough probes');
 end
 
-pos = regexp(probeHeader, params.keys, 'end');
+pos = regexp(probeHeader, params(1).keys, 'end');
 trimHeader = {};
 for n = 1:length(probeHeader)
     trimHeader{end+1} = probeHeader{n,1}(1:pos{n,1});
@@ -24,11 +24,12 @@
 indexTotal = zeros(length(trimHeader),1);
 for n = 1:length(uniqueHeader)
     index = ismember(trimHeader, uniqueHeader{n,1});
-    if sum(index) < params.number && checkSpecialTranscripts(uniqueHeader{n,1}) % for Bin's special sequences
+    if sum(index) < params(1).number &&...
+            checkSpecialTranscripts(uniqueHeader{n,1},params) % for Bin's special sequences
         indexTotal = indexTotal+index;
-        if params.verbose
+        if params(1).verbose
             disp(['  transcript ' uniqueHeader{n,1} ...
-                ' has less than ' num2str(params.number) ' probes']);
+                ' has less than ' num2str(params(1).number) ' probes']);
         end
     end
 end
@@ -43,29 +44,29 @@
 indexTotal = [];
 for n = 1:length(uniqueHeader)
     index = ismember(trimHeader, uniqueHeader{n,1});
-    if sum(index) > params.number
+    if sum(index) > params(1).number
         index=find(index>0);
-        index=index(params.number+1:end);
+        index=index(params(1).number+1:end);
         indexTotal=[indexTotal;index];
     end
 end
 
 probeHeader(indexTotal) = [];
 probeSequence(indexTotal) = [];
 
-disp('  randomizing and saving the list of probes');
-indexTotal = randperm(length(probeHeader))';
-probeHeader = probeHeader(indexTotal);
-probeSequence = probeSequence(indexTotal);
-probeList = [params.species '.probes.fas'];
+% disp('  randomizing and saving the list of probes');
+% indexTotal = randperm(length(probeHeader))';
+% probeHeader = probeHeader(indexTotal);
+% probeSequence = probeSequence(indexTotal);
+probeList = [params(1).species '.probes.nr.fas'];
 if exist(probeList, 'file')
     delete(probeList);
 end
 fastawrite(probeList,probeHeader,probeSequence);
 
 %% Check how many transcripts are left after this step of screening
-[geneNumLeft,geneNumDelete]=checkTranscriptsLeft(adapterList,probeHeader);
-if params.verbose
+[geneNumLeft,geneNumDelete]=checkTranscriptsLeft(adapterList,probeHeader,params);
+if params(1).verbose
     disp([num2str(geneNumDelete) ' out of ' num2str(geneNumLeft+geneNumDelete)...
         ' FISH escaped FISHerMan''s net']);
 end
diff --git a/main.m b/main.m
@@ -65,14 +65,14 @@
 
 %% Save the probes of each transcripts into individual files
 [probeHeader,probeSequence,probeSequence3Seg,probeSequenceCore]...
-    =blastOtherSteps(adapterList,probeHeader,probeSequence,probeSequence3Seg,probeSequenceCore);
+    =blastOtherSteps(adapterList,probeHeader,probeSequence,probeSequence3Seg,probeSequenceCore,params.otherSteps);
 
 %% Remove non-specific probes against the abundant rna database
 [probeHeader,probeSequence3Seg,probeSequence,probeSequenceCore]...
-    =blastAbundantRNA(adapterList,probeHeader,probeSequence3Seg,probeSequence,probeSequenceCore);
+    =blastAbundantRNA(adapterList,probeHeader,probeSequence3Seg,probeSequence,probeSequenceCore,params.arna);
 
 %% Generate the list of probes
-probeList=generateProbeList(adapterList,probeHeader,probeSequence);
+probeList=generateProbeList(adapterList,probeHeader,probeSequence,params.probeList);
 
 disp('done designing FISH probes');
 disp('FISHerMan is at rest');

diff --git a/oligosParse.m b/oligosParse.m
@@ -114,7 +114,7 @@
 indexTotal = zeros(length(trimNames),1);
 for n = 1:length(uniqueNames)
     index = ismember(trimNames, uniqueNames{n,1});
-    if sum(index) < params(1).number &&... 
+    if sum(index) < params(1).number &&...
             checkSpecialTranscripts(uniqueNames{n,1},params) % for Bin's special sequences
         indexTotal = indexTotal+index;
         if params(1).verbose

diff --git a/readParameters.m b/readParameters.m
@@ -23,6 +23,9 @@
 oligos=xmlParse(parameters, 'parameters', 'oligos');
 adapters=xmlParse(parameters, 'parameters', 'adapters');
 onePCR=xmlParse(parameters, 'parameters', 'onePCR');
+otherSteps=xmlParse(parameters, 'parameters', 'otherSteps');
+arna=xmlParse(parameters, 'parameters', 'arna');
+probeList=xmlParse(parameters, 'parameters', 'probeList');
 
 %% Parse general parameters
 verbose=xmlParse(general, 'general', 'verbose');
@@ -180,3 +183,60 @@
     'gf',char(gf.getFirstChild.getData),...
     'grr',char(grr.getFirstChild.getData));
 
+%% Parse parameters for otherSteps
+key1=xmlParse(otherSteps, 'otherSteps', 'key1');
+thres=xmlParse(otherSteps, 'otherSteps', 'thres');
+querySize=xmlParse(otherSteps, 'otherSteps', 'querySize');
+blastArgs1=xmlParse(otherSteps, 'otherSteps', 'blastArgs1');
+blastArgs2=xmlParse(otherSteps, 'otherSteps', 'blastArgs2');
+grr=xmlParse(otherSteps, 'otherSteps', 'grr');
+T7r=xmlParse(otherSteps, 'otherSteps', 'T7r');
+rRr=xmlParse(otherSteps, 'otherSteps', 'rRr');
+rGr=xmlParse(otherSteps, 'otherSteps', 'rGr');
+rBr=xmlParse(otherSteps, 'otherSteps', 'rBr');
+rIRr=xmlParse(otherSteps, 'otherSteps', 'rIRr');
+
+params.otherSteps = struct('species',species,...
+    'verbose',str2double(verbose.getFirstChild.getData),...
+    'keys',char(key1.getFirstChild.getData),...
+    'thres',str2double(thres.getFirstChild.getData),...
+    'querySize',str2double(querySize.getFirstChild.getData),...
+    'blastArgs1',char(blastArgs1.getFirstChild.getData),...
+    'blastArgs2',char(blastArgs2.getFirstChild.getData),...
+    'grr',char(grr.getFirstChild.getData),...
+    'T7r',char(T7r.getFirstChild.getData),...
+    'rRr',char(rRr.getFirstChild.getData),...
+    'rGr',char(rGr.getFirstChild.getData),...
+    'rBr',char(rBr.getFirstChild.getData),...
+    'rIRr',char(rIRr.getFirstChild.getData));
+
+%% Parse parameters for arna
+key1=xmlParse(arna, 'arna', 'key1');
+seqNum=xmlParse(arna, 'arna', 'seqNum');
+thres=xmlParse(arna, 'arna', 'thres');
+querySize=xmlParse(arna, 'arna', 'querySize');
+DbSize=xmlParse(arna, 'arna', 'DbSize');
+blastArgs=xmlParse(arna, 'arna', 'blastArgs');
+parallel=xmlParse(arna, 'arna', 'parallel');
+
+params.arna = struct('species',species,...
+    'verbose',str2double(verbose.getFirstChild.getData),...
+    'keys',char(key1.getFirstChild.getData),...
+    'seqNum',str2double(seqNum.getFirstChild.getData),...
+    'thres',str2double(thres.getFirstChild.getData),...
+    'querySize',str2double(querySize.getFirstChild.getData),...
+    'DbSize',str2double(DbSize.getFirstChild.getData),...
+    'blastArgs',char(blastArgs.getFirstChild.getData),...
+    'parallel',str2double(parallel.getFirstChild.getData));
+
+%% Parse parameters for probeList
+key1=xmlParse(probeList, 'probeList', 'key1');
+num=xmlParse(probeList, 'probeList', 'number');
+dir1=xmlParse(probeList, 'probeList', 'dir1');
+
+params.probeList = struct('species',species,...
+    'verbose',str2double(verbose.getFirstChild.getData),...
+    'keys',char(key1.getFirstChild.getData),...
+    'number',str2double(num.getFirstChild.getData),...
+    'specialTranscripts',char(dir1.getFirstChild.getData));
+