#14 - amend mmseqs parameters adjustment for nucleotide search

Author: endixk

src/module/ProfileModule.java

@@ -837,11 +837,11 @@ public static void run(String[] args) {
 					MMseqsSearchResultEntity res = MMseqsEasySearchProcess.search(
 							PathConfig.SeqPath + "nuc" + File.separator + "ITS.fa",
 							PathConfig.InputIsFolder ? PathConfig.InputPath + GenericConfig.FILENAME : PathConfig.InputPath,
-							PathConfig.TempPath, 500, 500, true, GenericConfig.ThreadPoolSize);
+							PathConfig.TempPath, 3, 500, 500, true, GenericConfig.ThreadPoolSize);
 					res.reduce();
 					ProfilePredictionEntity pp = MMseqsEasySearchProcess.parse(res);
 					res.remove();
-					if(pp.nseq() > 0) MMseqsEasySearchProcess.validate(pp, PathConfig.SeqPath + "nuc" + File.separator + "ITS.fa", PathConfig.TempPath, GenericConfig.ThreadPoolSize);
+					if(pp.nseq() > 0) MMseqsEasySearchProcess.validate(pp, PathConfig.SeqPath + "nuc" + File.separator + "ITS.fa", PathConfig.TempPath, 3, GenericConfig.ThreadPoolSize);
 					pps.add(pp);
 
 					if(pp.valid()) Prompt.print(ANSIHandler.wrapper("SUCCESS", 'g') + " : ITS sequence successfully extracted.");
@@ -1067,7 +1067,7 @@ public ProfilePredictionEntity call() {
 				updateProg(g, ANSIHandler.wrapper("V", 'R'), 1); printProg();
 				Prompt.talk(ANSIHandler.wrapper("[Phase 3 : Validation]", 'R'));
 				/* Run hmmsearch and find target gene among the predicted genes */
-				if(pp.nseq() > 0) MMseqsEasySearchProcess.validate(pp, PathConfig.SeqPath + dir + File.separator + cg + ".fa", PathConfig.TempPath, 1);
+				if(pp.nseq() > 0) MMseqsEasySearchProcess.validate(pp, PathConfig.SeqPath + dir + File.separator + cg + ".fa", PathConfig.TempPath, 0, 1);
 				/* Implementation note.
 				 *		Previously, contigs were removed right after AUGUSTUS finishes the prediction.
 				 *		However, to extract the cDNA sequence for the gene, contig should remain intact.

src/module/ProfileProModule.java

@@ -322,7 +322,7 @@ public ProfilePredictionEntity call() {
 				Prompt.talk(ANSIHandler.wrapper("[Phase 3 : Validation]", 'R'));
 				ProfilePredictionEntity pp = new ProfilePredictionEntity(cg, ProfilePredictionEntity.TYPE_PRO);
 				for(String seq : SEQS) pp.addSeq(seq);
-				MMseqsEasySearchProcess.validate(pp, PathConfig.SeqPath + "pro" + File.separator + cg + ".fa", PathConfig.TempPath, 1);
+				MMseqsEasySearchProcess.validate(pp, PathConfig.SeqPath + "pro" + File.separator + cg + ".fa", PathConfig.TempPath, 0, 1);
 
 				/* Record obtained result */
 				String result;

src/pipeline/UFCGMainPipeline.java

@@ -37,8 +37,8 @@
 @SuppressWarnings("restriction")
 public class UFCGMainPipeline {
 	public static final String VERSION = "1.0.5";
-	public static final Boolean STABLE = true;
-	public static final String RELEASE_DATE = "Aug 2023";
+	public static final Boolean STABLE = false;
+	public static final String RELEASE_DATE = "Sep 2023";
 	public static final String CITATION = " Kim, D., Gilchrist, C.L.M., Chun, J. & Steinegger, M. (2023)\n"
 										+ " UFCG: database of universal fungal core genes and pipeline for genome-wide phylogenetic analysis of fungi.\n"
 										+ " Nucleic Acids Research, 51(D1), D777-D784, doi:10.1093/nar/gkac894.\n";

src/process/MMseqsEasySearchProcess.java

@@ -65,12 +65,12 @@ public static String checkORF(ProfilePredictionEntity pp, String prtn, String or
 	static String TASK = null;
 	public static void setTask(String task) {TASK = task;}
 
-	public static MMseqsSearchResultEntity search(String queryPath, String targetPath, String tmpPath, int uoff, int doff, boolean abs, int threads) {
+	public static MMseqsSearchResultEntity search(String queryPath, String targetPath, String tmpPath, int searchType, int uoff, int doff, boolean abs, int threads) {
 		String resultPath = tmpPath + GenericConfig.SESSION_UID + "_" + targetPath.substring(targetPath.lastIndexOf(File.separator) + 1) + ".m8";
 
 		MMseqsWrapper mm = new MMseqsWrapper();
 		mm.setEasySearch(queryPath, targetPath, resultPath, tmpPath);
-		mm.setSearchType(0);
+		mm.setSearchType(searchType);
 		mm.setThreads(threads);
 		mm.exec();
 
@@ -90,12 +90,12 @@ public static MMseqsSearchResultEntity search(String queryPath, String targetPat
 
 		return res;
 	}
-	public static MMseqsSearchResultEntity search(String queryPath, String targetPath, String tmpPath, double evalue, int threads, double cov) {
+	public static MMseqsSearchResultEntity search(String queryPath, String targetPath, String tmpPath, int searchType, double evalue, int threads, double cov) {
 		String resultPath = tmpPath + GenericConfig.SESSION_UID + "_" + targetPath.substring(targetPath.lastIndexOf(File.separator) + 1) + ".m8";
 
 		MMseqsWrapper mm = new MMseqsWrapper();
 		mm.setEasySearch(queryPath, targetPath, resultPath, tmpPath);
-		mm.setSearchType(0);
+		mm.setSearchType(searchType);
 		mm.setEvalue(evalue);
 		mm.setCoverage(cov);
 		mm.setThreads(threads);
@@ -124,16 +124,16 @@ public static ProfilePredictionEntity parse(MMseqsSearchResultEntity res) {
 		return pp;
 	}
 
-	public static void validate(ProfilePredictionEntity pp, String seqPath, String tmpPath, int threads) {
+	public static void validate(ProfilePredictionEntity pp, String seqPath, String tmpPath, int searchType, int threads) {
 		String queryPath = pp.export();
 
 		// 3-step coverage search
-		MMseqsSearchResultEntity res = search(queryPath, seqPath, tmpPath, GenericConfig.EvalueCutoff, threads, 0.8);
+		MMseqsSearchResultEntity res = search(queryPath, seqPath, tmpPath, searchType, GenericConfig.EvalueCutoff, threads, 0.8);
 		if(res.size() == 0 && GenericConfig.SENS > 1) {
-			res = search(queryPath, seqPath, tmpPath, GenericConfig.EvalueCutoff, threads, 0.5);
+			res = search(queryPath, seqPath, tmpPath, searchType, GenericConfig.EvalueCutoff, threads, 0.5);
 		}
 		if(res.size() == 0 && GenericConfig.SENS > 2) {
-			res = search(queryPath, seqPath, tmpPath, GenericConfig.EvalueCutoff, threads, 0.0);
+			res = search(queryPath, seqPath, tmpPath, searchType, GenericConfig.EvalueCutoff, threads, 0.0);
 		}
 
 		res.assignLocs();