Difference between revisions of "Gaucher Disease: Task 02 - Alignments"

From Bioinformatikpedia
(Comparison)
(T-Coffee)
Line 331: Line 331:
 
* Results for T-Coffee of : [[Multiple_Sequence_Alignment:_T-Coffee_-_set_3|Set3 ]]
 
* Results for T-Coffee of : [[Multiple_Sequence_Alignment:_T-Coffee_-_set_3|Set3 ]]
   
  +
The Alignment of '''set 1''' generated by T-Coffee is similar to its alignment made by Muscle. There are again no conserved columns because of the shift of the sequence fragments. The T-Coffee alignment has a few more gaps, which are also very long, and cause a different fragment shift than in the Muscle alignment. Interestingly the short sequences A9UD35 and D1L2S0 have one gap. This extremely long contigous gap shifts the last amino acid of the sequence to the last position of the alignment. As these two amino acids are not identical to the last amino acids of the remaining sequences, there seem to be no reason for this observation.
 
   
   
Line 349: Line 349:
 
|-
 
|-
 
|Number of contiguous gaps in P04062
 
|Number of contiguous gaps in P04062
|
+
|0
 
|
 
|
 
|
 
|

Revision as of 14:30, 6 May 2013

Work is still in progress here. Please check today evening ;)

Alignments allow a comparisons of Strings. In the field of bioinformatics, sequence alignments show the relation between two or more sequences.

Theoretical Background

Description

Pairwise or multiple alignments often contain an aditional line below the proper alignment (or between two aligned sequences for pairwise alignments). This line gives a more accurate description of the relation of the aligned residues above. The symbols show if there is match between identical amino acids or if they are only similar.

Symbols for describing sequence alignments
Program(s) Symbol Example Meaning
MSAs * blub identical residues
: similar residues
.
(Psi-)BLAST same letter A
A
A
identical residues
+ L
+
V
similar residues
HHblits | AF
| |
AW
identical or very similar residues?
+ T
+
S
similar residues?
. N
.
H
non-similar residues?

Sequence Searches

Sequence searches with our query protein sequence, P04062.fasta, were done with the following programs:

  • BLAST
    • using standard parameters
    • against big_80
    • against big
  • Psi-BLAST
    • with number of shown hits and alignments set to 10000 (-b, -v options), so that all the hits will be shown.
    • with all combinations of:
      • 2 iterations: 1 iterations against big_80 followed by 1 iteration against big
      • 10 iterations: 9 iterations against big_80 followed by 1 iteration against big
      • default E-value cutoff (0.002)
      • E-value cutoff 10E-10
    • other options leaved default
  • HHblits
    • with number of shown hits and alignments set to 10000 (-Z, -B options), as in Psi-BLAST
    • with all combinations of:
      • 2 iterations against uniprot_20
      • 10 iterations against uniprot_20
      • default E-value cutoff (0.002)
      • E-value cutoff 10E-10
    • other options leaved default

The script run.pl was written and used for the runs. PSSM files - a3m and hhr for HHblits, chk ("checkpoint") and PSSM for Psi-BLAST were created in order to start the search against another database, from big_80 to big for Psi-BLAST and later against a PDB database for the evaluation.

For Psi-BLAST, first a search against big_80 was done in order to create a good profile, then a last iteration against big was done with this profile. The idea was to get as many hits as possible, so that the results will be comparable with HHblits, where the runs were made against the clustered HMM database. All uniprot_20 cluster members were count in the following comparison and evaluation.

Comparison

The results were parsed and analysed using the script parse_output.pl.

Number and overlap of hits Number of found hits with each program and parameter combination and overlap of hits between some are summerized in the following tables. <todo tables>

Furthermore, E-value, percentage identity and alignment length distributions were plottet.

E-value distribution

E-value distribution of HHblits hits (2 iterations against uniprot_20 with default E-value cutoff, 0.001), Psi-BLAST hits (1 iteration against big_80 followed by one iteration against big with default E-value cutoff, 0.002) and BLAST hits (against big with default parameters). On the X-axis -log of the E-value is plotted, so the smallest (the best) E-value is on the right side. On the Y-axis is logarithmic frequency of the values on the X-axis.

E-value distribution of HHblits hits (2 iterations against uniprot_20 with E-value cutoff 10E-10) and Psi-BLAST hits (1 iteration against big_80 followed by one iteration against big with E-value cutoff 10E-10). On the X-axis -log of the E-value is plotted, so the smallest (the best) E-value is on the right side. On the Y-axis is logarithmic frequency of the values on the X-axis.

E-value distribution of HHblits hits (10 iterations against uniprot_20 with default E-value cutoff, 0.001) and Psi-BLAST hits (9 iteration against big_80 followed by one iteration against big with default E-value cutoff, 0.002). On the X-axis -log of the E-value is plotted, so the smallest (the best) E-value is on the right side. On the Y-axis is logarithmic frequency of the values on the X-axis.

E-value distribution of HHblits hits (10 iterations against uniprot_20 with E-value cutoff 10E-10) and Psi-BLAST hits (9 iteration against big_80 followed by one iteration against big with E-value cutoff 10E-10). On the X-axis -log of the E-value is plotted, so the smallest (the best) E-value is on the right side. On the Y-axis is logarithmic frequency of the values on the X-axis.

Percentage identity distribution


Alignment length distribution

Evaluation

Validation against COPS L30 - L60 groups was made.

Multiple Sequence Alignment

For the multiple sequence alignments three sets were created. Therefore the results of the previous task were according to their sequnece identiy to the native protein sequence of glucocerebrosidase. Set 1 and set 2 contains 10 sequences, including the native protein sequence to keep the alignments in relation to the Gaucher's disease causing protein. For the remainig 9 sequences have a sequence identity to the glucocerebrosidase of more than 60% in set 1 and less than 30% in set 2. In set 3, the sequences are over the whole range of sequence identity. The multiple sequence alignments were made with Clustalw, Muscle and T-Coffee.

Set 1: sequence identity >60%
ID Identity in %
P04062 100
A9UD35 84.1
D1L2S0 83.0
3gxi_A 99.8
2nt1_A 99.8
F6WDY8 90.7
Q2KHZ8 89.2
F5CB27 81.8
F5H241 98.2
B7Z6S9 99.8
Set 2: sequence identity <30%
ID Identity in %
P04062 100
H6CEV7 26.5
Q21GD0 24.3
I1WBF3 23.7
I9HH59 29.4
D0TN48 25.4
B1VPJ0 27.5
E2LY19 24.8
K9HBW2 25.9
B5QQZ8 27.1
Set 3: sequence identity over all
ID Identity in % ID Identity in %
P04062 100 B5DYA3 32.0
B4JTN5 34.1 F5CB37 81.8
E1ZYU8 42.7 J2STU0 34.2
F4E6W5 26.4 G9BHQ3 80.7
F4X220 41.7 2nsx_B 99.8
E5UPZ0 21.6 G9BHQ5 82.5
C6A5Q0 25.0 D0TIX6 25.8
B1QKT0 27.8 1y7v_A 99.8
D4SV88 34.6 A9UD58 80.4
A9UD54 81.8 G9MUP6 20.7

ClustalW

Multiple alignments generated with ClustalW:

  • Results for Clustalw of  : Set1
  • Results for Clustalw of  : Set2
  • Results for Clustalw of  : Set3


Set 1, which includes sequences with high similarity has only 11 conserved columns. But these columns lie densly in an area of 55 amio acids. Also columns that are not conserved by identical residues, but have similar amino acids can be found in this high conserved region (marked blue in the linked multiple alignment results above). The gaps of the alignment split the protein into 7, mostly longer parts of of residues. Especially the last 400 residues have a very high sequence identity between the clucosylcerebrosidase and 6 other proteins.

Set 2 has less conserved columns than set 1. They are spread over the alignment and build no conserved region. The contiguous gaps are more but shorter.

There exist no conserved columns for set 3. This could result from the greater number of sequences in the set. The more sequences are in an alignment the lesser the probability of having conserved columns. Because of the sequences having an identity over the whole range, the sequences with a low identity cause to this loss of a conserved region. This can be seen by looking only on the sequences with high similarity (marked green in the alignment of set 3).


Results of ClustalW in numbers
Set 1 Set 2 Set 3
Conserved Columns 11 8 0
Number of contiguous gaps in P04062 7 18 18

Muscle

Multiple alignments generated with Muscle:

  • Results for Muscle of  : Set1
  • Results for Muscle of  : Set2
  • Results for Muscle of  : Set3

In Set 1 the native protein sequence itself has no gap. There are only gaps at the beginning of the sequence, because the length of the alignment is longer than the length of clucocerebrosidase sequence. There also exist no conserved columns. This is caused because of the shift of the sequences. So, in no alignment position, the residues of all sequences are aligned. The whole alignment has only one long contigous gap inside of the protein sequence F5H241. Some of the shorter sequences (D1L2S0, A9UD35, F5CB27) are aligned at these alignment positions, where F5H241 has its gap. If only the aligned residues would be considered and the gaps were neglected, there would be a lot of conserved columns.

In contrary to set 1, set 2 has 10 conserved columns. However they are widly spread over the alignment. The same observation can be made of columns with similar residues. Through this scattering it seems rather a randomly generated conserved column than a conservation due to functionaly reasons. The high number as well as the partly great length of the contigous gaps straighten the alignment. The alignment gives a great overview of the relation between sequences with low similarity and sequence identity. It also shows that for finding functional important areas that are conserved, a higher similarity and identity is needed.

Set 3 has also no conserved columns, for the same reason as explained befor, for set 1. It has indeed the highest number of gaps of the three sets. But contrary to the alignment of set 1 the contigous gaps are shorter, partially they are only 1-2 gaps long. So the alignment appears less straightened, as the sequences with high sequence identity keep the alignment a bit compact.


Results of Muscle in numbers
Set 1 Set 2 Set 3
Conserved Columns 0 10 0
Number of contiguous gaps in P04062 1 27 36

T-Coffee

Multiple alignments generated with T-Coffee:

  • Results for T-Coffee of  : Set1
  • Results for T-Coffee of  : Set2
  • Results for T-Coffee of  : Set3

The Alignment of set 1 generated by T-Coffee is similar to its alignment made by Muscle. There are again no conserved columns because of the shift of the sequence fragments. The T-Coffee alignment has a few more gaps, which are also very long, and cause a different fragment shift than in the Muscle alignment. Interestingly the short sequences A9UD35 and D1L2S0 have one gap. This extremely long contigous gap shifts the last amino acid of the sequence to the last position of the alignment. As these two amino acids are not identical to the last amino acids of the remaining sequences, there seem to be no reason for this observation.


Results of Muscle in numbers
Set 1 Set 2 Set 3
Conserved Columns 0
Number of contiguous gaps in P04062 0

Alignment Comparison