Problem set 11 Bruzzese + Inskeep +Nemera

Exercise11_files/Problem1/Problem1script.sh

@@ -0,0 +1,16 @@
+for i in *ref
+do
+    ../../../../muscle.exe -in $i -out $i.aln
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmbuild $i.hmm $i.aln
+done
+
+
+for i in *fasta
+do
+/home/Mati/hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout $i.sigma.hits sigma.ref.hmm $i
+/home/Mati/hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout $i.sporecoat.hits sporecoat.ref.hmm $i                   
+/home/Mati/hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout $i.transporter.hits transporter.ref.hmm $i
+done
+
+cat *.hits | grep "tr|" | sed -E 's/[tr|A-Z0-9]+\_9//' | awk '{print $1 " " $3 " " $5}' > bacteriahmmout.txt
+

Exercise11_files/Problem1/exercise11.py

@@ -0,0 +1,14 @@
+###Question 1###
+
+for i in *ref
+do
+    ../../../../muscle.exe -in $i -out $i.aln
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmbuild $i.hmm $i.aln
+done
+
+for i in *fasta
+do
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout $i.sigma.hits sigma.ref.hmm $i
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout $i.sporecoat.hits sporecoat.ref.hmm $i
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout $i.transporter.hits transporter.ref.hmm $i
+done

Exercise11_files/exercise11.py

@@ -0,0 +1,14 @@
+###Question 1###
+
+for i in *ref
+do
+    ../../../../muscle.exe -in $i -out $i.aln
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmbuild $i.hmm $i.align
+done
+
+for i in *fasta
+do
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout sigma.hits sigma.hmm $i
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout sporecoat.hits sporecoat.hmm $i
+    ../../../../hmmer-3.1b2-cygwin64/binaries/hmmsearch --tblout transporter.hits transporter.hmm $i
+done

exercise11_problem2.py

@@ -0,0 +1,51 @@
+import re
+# unzip files in unix
+# open fastafile
+thefile=open('./Exercise11_files/Problem2/motifsort.fasta',"r")
+
+# search strings
+Motif1= r'AKKPRVZE'
+Motif2 = r'AAQWWRNYGG'
+
+# lists!
+m1_id = []
+m1_seq = []
+
+m2_id = []
+m2_seq = []
+
+other_id = []
+other_seq = []
+
+# loop over the file
+for line in thefile:
+    line = line.strip()
+    if ">" in line:
+        seqid = line
+    elif re.search(Motif1, line):
+        m1 = line
+        m1_id.append(seqid)
+        m1_seq.append(m1)
+    elif re.search(Motif2, line):
+        m2 = line
+        m2_id.append(seqid)
+        m2_seq.append(m2)
+    else:
+        other_id.append(seqid)
+        other_seq.append(line)
+
+# write fasta files from lists :)
+ofile = open("m1_motif.txt", "w")
+for i in range(len(m1_id)):
+    ofile.write(m1_id[i] + "\n" +m1_seq[i] + "\n")
+ofile.close()
+
+ofile = open("m2_motif.txt", "w")
+for i in range(len(m2_id)):
+    ofile.write(m2_id[i] + "\n" +m2_seq[i] + "\n")
+ofile.close()
+
+ofile = open("other.txt", "w")
+for i in range(len(other_id)):
+    ofile.write(other_id[i] + "\n" +other_seq[i] + "\n")
+ofile.close()