ExPairwiseAlignment-AnswersEng

Answers to pairwise alignment exercise (English version)

• The exercise itself is here: ExPairwiseAlignment
• The Danish version of the answers is here: ExPairwiseAlignment-Answers

Part 0

Q0.1

Yes, the matrix has the same values in the cells as the one from the lecture.

Q0.2

The numbers are different, because the substitution matrices are not identical, but the resulting alignment is the same.

Part 1

Q1

FASTA format.

Q2.1

Alignment Score: 860.50

Length: 361, Identities: 176/361 (48.8%), Similarities: 214/361 (59.3%), Gaps: 92/361 (25.5%)

Seq1         1  ------------------------------------------------------------    0
                                                                            
Seq2         1  MRQSLKVMVLSTVALLFMANPAAASEEKKEYLIVVEPEEVSAQSVEESYDVDVIHEFEEI   60

Seq1         1  --------------------------------AQSVPWGISRVQAPAAHNRGLTGSGVKV   28
                                                :|:||||||.:....|||||:.|:|.:|
Seq2        61  PVIHAELTKKELKKLKKDPNVKAIEKNAEVTISQTVPWGISFINTQQAHNRGIFGNGARV  120

Seq1        29  AVLDTGISTHPDLNIRGGASFVPGEPSTQDGNGHGTHVAGTIAALNNSIGVLGVAPSAEL   88
                |||||||::||||.|.|||||:..|||..|.|||||||||||||||||||||||||||:|
Seq2       121  AVLDTGIASHPDLRIAGGASFISSEPSYHDNNGHGTHVAGTIAALNNSIGVLGVAPSADL  180

Seq1        89  YAVKVLGASGSGSVSSIAQGLEWAGNNGMHVANLSLGSPSPSATLEQAVNSATSRGVLVV  148
                ||||||..:||||::|:|||:|||.||.||:.|:||||.|.|:|||.|||.|.:.|:|:|
Seq2       181  YAVKVLDRNGSGSLASVAQGIEWAINNNMHIINMSLGSTSGSSTLELAVNRANNAGILLV  240

Seq1       149  AASGNSGAGSISYPARYANAMAVGATDQNNNRASFSQYGAGLDIVAPGVNVQSTYPGSTY  208
                .|:||:|...::|||||:..|||.|.|||..|||||.||..::|.||||||.|||.|:.|
Seq2       241  GAAGNTGRQGVNYPARYSGVMAVAAVDQNGQRASFSTYGPEIEISAPGVNVNSTYTGNRY  300

Seq1       209  ASLNGTSMATPHVAGAAALVKQKNPSWSNVQIRNHLKNTATSLGSTNLYGSGLVNAEAAT  268
                .||:|||||||||||.|||||.:.||::|.|||..:..|||.|||.:|||:|||:|..||
Seq2       301  VSLSGTSMATPHVAGVAALVKSRYPSYTNNQIRQRINQTATYLGSPSLYGNGLVHAGRAT  360

Seq1       269  R  269
                :
Seq2       361  Q  361

Q2.2

The alignment score increased to 916, all other values are unchanged, and the alignment is identical.

Q3

Alignment Score: 916.00

Length: 269, Identities: 176/269 (65.4%), Similarities: 214/269 (79.6%), Gaps: 0/269 (0.0%)

Seq1         1  AQSVPWGISRVQAPAAHNRGLTGSGVKVAVLDTGISTHPDLNIRGGASFVPGEPSTQDGN   60
                :|:||||||.:....|||||:.|:|.:||||||||::||||.|.|||||:..|||..|.|
Seq2        93  SQTVPWGISFINTQQAHNRGIFGNGARVAVLDTGIASHPDLRIAGGASFISSEPSYHDNN  152

Seq1        61  GHGTHVAGTIAALNNSIGVLGVAPSAELYAVKVLGASGSGSVSSIAQGLEWAGNNGMHVA  120
                ||||||||||||||||||||||||||:|||||||..:||||::|:|||:|||.||.||:.
Seq2       153  GHGTHVAGTIAALNNSIGVLGVAPSADLYAVKVLDRNGSGSLASVAQGIEWAINNNMHII  212

Seq1       121  NLSLGSPSPSATLEQAVNSATSRGVLVVAASGNSGAGSISYPARYANAMAVGATDQNNNR  180
                |:||||.|.|:|||.|||.|.:.|:|:|.|:||:|...::|||||:..|||.|.|||..|
Seq2       213  NMSLGSTSGSSTLELAVNRANNAGILLVGAAGNTGRQGVNYPARYSGVMAVAAVDQNGQR  272

Seq1       181  ASFSQYGAGLDIVAPGVNVQSTYPGSTYASLNGTSMATPHVAGAAALVKQKNPSWSNVQI  240
                ||||.||..::|.||||||.|||.|:.|.||:|||||||||||.|||||.:.||::|.||
Seq2       273  ASFSTYGPEIEISAPGVNVNSTYTGNRYVSLSGTSMATPHVAGVAALVKSRYPSYTNNQI  332

Seq1       241  RNHLKNTATSLGSTNLYGSGLVNAEAATR  269
                |..:..|||.|||.:|||:|||:|..||:
Seq2       333  RQRINQTATYLGSPSLYGNGLVHAGRATQ  361

Since the two sequences are of different length (see also the answer to next question), one could say that it makes the most sense to use the Smith-Waterman algorithm ("local alignment"); this allows the analysis of the differences and similarities to focus on the parts of the sequences which are actually comparable.

However, by using the global alignment alone one can easily see that the sequences are very similar, apart from the missing piece of approximately 90 amino acids at the start. So in this case, we have learned something extra about the sequences by first making a global alignment.

When two sequences are very similar, as is the case here, there is generally not much difference in the information you get by using local or global alignment.

Q4

• For P29600, the sequence is derived from the 3D structure. For P41363, the sequence is translated from DNA + information from the protein sequencing.

• SUBCELLULAR LOCATION: "Secreted protein" (for both).

• P29600 starts directly with the sequence of the mature protein. P41363 starts with a signal peptide (positions 1-24), then the pro-peptide (25-93), and then comes the mature protein. Note, that both the signal peptide (function:signal to the export of the protein) and pro-peptide (function: helps protein with to fold correctly) are removed from the "mature" protein. The difference is that P41363 is (mostly) translated from the DNA and therefore contains information from the entire coding sequence, whereas P29600 is derived from the 3D structure, which contains only the mature sequence. Immature Savinase actually does contain both a signal peptide and a pro-peptide (as can be dug up in the databases).

Q5

Pros: Same type protease (serine protease, S8 family). Thermostable (!). The protein is very similar to the Savinase at the sequence level.

Potential problems: High pH optimum, but this could possibly be optimized in the laboratory.

Part 2

Q6

# Length: 1255
# Identity: 110/1255 (8.8%)
# Similarity: 154/1255 (12.3%)
# Gaps: 992/1255 (79.0%)
# Score: -244.0

Note: negative score! (Alignment not shown)

Q7

# Length: 1290
# Identity: 73/1290 (5.7%)
# Similarity: 131/1290 (10.2%)
# Gaps: 1062/1290 (82.3%)
# Score: 158.5

(Alignment not shown. But this time, it looks completely different from the one made with end gap penalties.)

Q8

# Length: 296
# Identity: 71/296 (24.0%)
# Similarity: 129/296 (43.6%)
# Gaps: 73/296 (24.7%)
# Score: 173.0

Seq1         23  GSGVKVAVLDTGISTHPDLNIRGGASFVPGEPSTQDGNGHGTHVAGTIAA-------LNN    75
                 ||.....:|:..::.:.|.|:   .|.|      ..|..|||||| :|||       ..|
Seq2        234  GSFGTAEMLNYSVNIYDDGNL---LSIV------TSGGAHGTHVA-SIAAGHFPEEPERN   283

Seq1         76  SIGVLGVAPSAELYAVKV------LGASGSGSVSSIAQGLEWAGNNGMHVANLSLGSPS-   128
                      ||||.|::.::|:      ...:|:|.:.::.:.:    |:...:.|.|.|..: 
Seq2        284  -----GVAPGAQILSIKIGDTRLSTMETGTGLIRAMIEVI----NHKCDLVNYSYGEATH   334

Seq1        129  --PSATLEQAVNSAT-SRGVLVVAASGNSG--AGSISYP-ARYANAMAVGATDQNN----   178
                   .|..:.:.:|.|. ...::.|:::||:|  ..::..| ...::.:.|||....:    
Seq2        335  WPNSGRICEVINEAVWKHNIIYVSSAGNNGPCLSTVGCPGGTTSSVIGVGAYVSPDMMVA   394

Seq1        179  ----------NRASFSQYGAGLDIVAPGVNVQSTYPGSTYAS-----------LNGTSMA   217
                           |:.::|..|...| .|.||::.:  ||...||           :|||||:
Seq2        395  EYSLREKLPANQYTWSSRGPSAD-GALGVSISA--PGGAIASVPNWTLRGTQLMNGTSMS   451

Seq1        218  TPHVAGAAALV----KQKNPSWSNVQIRNHLKNTATSLGSTNLY--GSGLVNAEAA   267
                 :|:..|..||:    |..|..::...:|..|:|||....:..::  |.|::..:.|
Seq2        452  SPNACGGIALILSGLKANNIDYTVHSVRRALENTAVKADNIEVFAQGHGIIQVDKA   507

Q9

It is clear from the local alignment and from the global alignment without end gap penalties that the prokaryotic protease matches only a single area in the middle of the human protease. In contrast, this is not clear in the global alignment with end gap penalties, which "spreads" the short sequence over the entire length.

Note that the global alignment without end gap penalties can be regarded as a kind of compromise between global and local alignment.

For distantly related sequences, it would be best to use local alignment; this would actually provide an optimal analysis of the comparable part of the sequences.

Q10

Your answers will of course vary randomly, but generally I would expect a reply within these ranges:

# Length: 100-300 
# Identity: 20% -30%  
# Similarity: 30% -40%  
# Gaps #: 25% -40%  
# Score: 40-70 

So this is data from the local alignments you get to compare non-related sequences with the given length and amino acid composition.

The idea here of making Savinase / shuffled alignments is to get a "null model" that can be compared with the real Savinase / Human peptidase alignment. If you had completed the experiment 100 times instead of 3, you could have done statistics on the outcome and calculated confidence limits and therefore evaluated the degree of statistical significance from a given alignment score (more on statistical significance when we come to BLAST).

Q11

When we compare our Savinase / Human peptidase alignment (score: 173) with the "deliberately bad" Savinase / shuffled alignments, it does not seems so bad anymore. The score is clearly higher than what we got with the shuffled sequences. However, notice that one needs to look at the score to see a clear difference; the other values (identity, similarity) may be similar between the original alignment and the shuffled alignments.

As we will see when we learn about BLAST, the idea is to evaluate an alignment score against a reference of scores from unrelated sequences.

Part 3

Q12

BLOSUM90:
# Length: 279
# Identity: 73/279 (26.2%)
# Similarity: 107/279 (38.4%)
# Gaps: 91/279 (32.6%)
# Score: 147.5
BLOSUM30:
# Length: 326
# Identity: 76/326 (23.3%)
# Similarity: 149/326 (45.7%)
# Gaps: 88/326 (27.0%)
# Score: 342.5

Note how a matrix of a lower BLOSUM-value results in a longer local alignment with a lower % identity.

Q13

# Length: 1252
# Identity: 192/1252 (15.3%)
# Similarity: 228/1252 (18.2%)
# Gaps: 1006/1252 (80.4%)
# Score: 895,576

Note how the sequences are stretched out each time the amino acids are not similar.

This alignment does not provide any biological insight. If the gap penalty is low enough, any alignment gives a high score.

Epilogue

Q14

The sequence of GLB7A_CHITH that corresponds to the 6 positions long gap in GLBE_CHITH is "ALIGNE"