USalign - sequence dependent alignment

[martingb]

Dears,

First of all, thanks for all your work on developing software for structural alignments.

I have been using US-align (Version 20241108) guided by sequence alignment as:

Code: Select all

$ ./USalign 1.pdb.gz 2.pdb.gz -infmt1 0 -infmt2 0 -mol RNA -het 1 -seq -outfmt -1 -o <sup-pathfile>

And I have read the paper, documentation (USalign -h) and website but still can't correctly interpret some outputs.

Please, could you explain to me line by line how to interpret the results with the following output?

Code: Select all

 0 | Name of Structure_1: 1.pdb.gz:? (to be superimposed onto Structure_2)
 1 | Name of Structure_2: 2.pdb.gz:?
 2 | Length of Structure_1: 16 residues
 3 | Length of Structure_2: 15 residues
 4 |
 5 | User-specified initial alignment: TM/Lali/rmsd = 0.25858,   15,  3.018
 6 | Aligned length= 15, RMSD=   3.02, Seq_ID=n_identical/n_aligned= 1.000
 7 | TM-score= 0.23648 (normalized by length of Structure_1: L=16, d0=0.50)
 8 | TM-score= 0.21836 (normalized by length of Structure_2: L=15, d0=0.40)
 9 | (You should use TM-score normalized by length of the reference structure)
10 |
11 | (":" denotes residue pairs of d < 5.0 Angstrom, "." denotes other aligned residues)
12 | aaggagguaaaauucu
13 |  .......::::::::
14 | -aggagguaaaauucu
15 |
16 | #Total CPU time is  0.00 seconds

These are some questions that come to me:

• How should I read the standard output lines for -outfmt -1?

• Why are there 2 values for RMSD, Aligned length and TMscore (line 5 and 6)?

• In line 5 which starts with "User-specified initial alignment: TM/Lali/rmsd..."
  • Are those the values that I am looking for?
  • Are the RMSD values in lines 5 and 6 the same?
  • Why do they have different significant decimals among them and compared to lines 6, 7 and 8?
  • Which target sequence length is considered to calculate that 'TM' value (assume TMscore)?
  • The 'TM' value in line 5 shouldn't be one of line 7 or 8?

Best,
Martín

[zcx@umich.edu]

1) How should I read the standard output lines for -outfmt -1?

For easy of parsing, you may use -outfmt 1 or -outfmt 2, which are more compact.

2) Why are there 2 values for RMSD, Aligned length and TMscore (line 5 and 6)?
In line 5 which starts with "User-specified initial alignment: TM/Lali/rmsd..."
Are those the values that I am looking for?
Are the RMSD values in lines 5 and 6 the same?

Yes, when using -I or -seq, RMSD at line 5 and line 6 are the same, just different number of significant decimals.
On the other hand, when using -i, RMSD at line 5 and line 6 may be different.

3) Why do they have different significant decimals among them and compared to lines 6, 7 and 8?
Which target sequence length is considered to calculate that 'TM' value (assume TMscore)?
The 'TM' value in line 5 shouldn't be one of line 7 or 8?

TM-score at line 7 and 8 are different because it is normalized by the sequence length of 1.pdb.gz and 2.pdb.gz, respectively.

The TM-score at line 5 is no less than that of line 7 and 8 because it is normalized by the alignment length rather than the length 1.pdb.gz and 2.pdb.gz.

[martingb]

Thank you for your answer but I still have some doubts...

• Are the 'Lali' (line 5) and 'Aligned_length' (line 6) values meaning the same?

• Knowing that TM-score at line 5 is normalized by alignment length (Lali = 15, line 5), which is the same value of Structure 2 length (15 residues, line 3 and L = 15, line 8), shouldn't both TM-scores (line 5 and line 8) be the same? Why are different?

[otiscavin]

USalign is a structural alignment tool for macromolecules (proteins, nucleic acids, or complexes) developed as part of the TM-align family (by Yang Zhang’s group, University of Michigan). Its main purpose is to compare 3D structures of biomolecules, producing an optimal alignment and similarity score. Sprunki

[ningfieldin]

USalign is a structural alignment tool for macromolecules (proteins, nucleic acids, or complexes) developed as part of the TM-align family (by Yang Zhang’s group, University of Michigan). Its main purpose is to compare 3D structures of biomolecules, producing an optimal alignment and similarity scoreSpace Waves.

I totally agree with you