I tried to model a protein sequence (4 chains) using RosettaCM. The structure of the protein is already available in PDB (2ERJ). The issue I am facing - the final modelled structure has all the 4 chains very far apart, unlike the pdb. What exactly is going wrong? 

I am obtaining the alignment file from Clustal omega. The target fasta sequence is given as chain1/chain2/chain3/chain4 and the 2ERJ fasta is also given in the same format in clustal omega so that the final alignment has only 2 unique identifiers in the ali file, like this:

2ERJ1|Chains      GMLSLELCDDDPPEIPHATFKAMAYKEGTMLNCECKRGFRRIKSGSLYMLCTGSSSHSSW     60
IL2Ra              -----ELCDDDPPEIPHATFKAMAYKEGTMLNCECKRGFRRIKSGSLYMLCTGNSSHSSW     55
                        ************************************************.******

2ERJ1|Chains      DNQCQCTSSATRSTTKQVTPQPEEQKERKTTEMQSPMQPVDQASLPGHCREPPPWENEAT     120
IL2Ra              DNQCQCTSSATRNTTKQVTPQPEEQKERKTTEMQSPMQPVDQASLPGHCREPPPWENEAT     115
                   ************.***********************************************

2ERJ1|Chains      ERIYHFVVGQMVYYQCVQGYRALHRGPAESVCKMTHGKTRWTQPQLICTGEMETSQFPGE     180
IL2Ra              ERIYHFVVGQMVYYQCVQGYRALHRGPAESVCKMTHGKTRWTQPQLICTGEMETSQFPGE     175
                   ************************************************************

2ERJ1|Chains      EKPQASPEGRPESETSCLVTTTDFQIQTEMAATMETSTGHHHHH----------------     224
IL2Ra              EKPQASPEGRPESETSCLVTTTDFQIQTEMAATMETSIFTTEYQVAVAGCVFLLISVLLL     235
                   *************************************    .::                

and so on.

I used the following commands:

~wholepath/tools/protein_tools/scripts/setup_RosettaCM.py --fasta IL2Ra.fasta --alignment alignmentfile.aln --alignment_format clustalw --templates 2ERJ.pdb

~wholepath/main/source/bin/rosetta_scripts.default.linuxgccrelease @flags -database /wholepath/main/database -nstruct 10

What is it that I am doing wrong which is leading to all the modelled chains being very far apart in the final structures? 

Kindly help.