Science -II (Biology)

Final Exam April 2022

Roll No.: Seat No.:

Date of Exam: Investigators Signature:
Duration: 3hrs
Total Marks: 90

Note: Answer PART-A in the space provided, and use additional sheet for PART-B.

PART-A: (Marks: 32)

1. Show how a self-dot plot looks like if the sequence contains a (i) tandem repeat and (ii) an
inverted repeat. [2]
Soln:

2. Echolation in sperm whale is an ancient or derived characteristic? How was it determined?

[2]
Soln: Echolation is an ancient characteristic in sperm whale. Analysis of the two mitochondrial
ribosomal gene suggested that sperm whales are more closely related to baleen whales than to
other toothed whales. This implies echolocation is present in both branches of whales, which
suggests that the common ancestor had this characteristic which was subsequently lost in the
baleen whales as a result of adaptation to their lifestyle. Thus echolation is an ancient
characteristic in sperm whale.

3. Scores of ungapped local alignment of random sequences follow ---------- distribution. [1]
Soln: extreme-value

4. (a) What’s the underlying assumption in the computation of sum of pairs scoring method in
MSA? [2]
(b)Using the scoring scheme (2, -1, -2), compute the score for the column:

()
G
A
−
G
Soln: (a)All the residues in a column represent the ancestral residues as well as recent
evolutionary changes. Since any one of the residues could have been the ancestor residue,
the computation of sum of pairs in MSA by scoring every residue with every other residue in
each column is carried assuming each one as the ancestor residue.
(b) -1-2+1-2-1-2 = -6

5. Is e-value in the BLAST program dependent on the size of database being searched? If yes,
will it increase/decrease with increase in the size of the database? [2]
Soln: Yes. It will decrease.

6. How would you randomize a nucleotide sequence of length L = 100 without changing the
composition of the nucleic acids? [2]
Soln: (a) In a loop over a very large N (~ 10 times L), call two random numbers between (1 to L)
and swap the positions of the two amino acids. R.
Else, Compute the frequencies of each of the nucleic acids in the given sequence. Break the
interval of 0 – 1 into parts corresponding to these frequencies. Call a random no. between 0 – 1.
Depending on the range in which the random no. falls, add that particular nucleic acid. Repeat
this 100 times.

7. Give one example of biological relevance for analyzing the number and distribution of k-
tuples: (i) k=1, (ii) k=2, (iii)k=3?
[3]
Soln: (i) k=1: composition of the sequence, origin of replication, melting point of DNA.
(ii) k=2: distribution of dinucleotide GC – helps in identifying protein-coding regions, set of
di-nucleotide relative abundance values constitutes a “genomic signature” of an organism,
Useful in identifying genomic islands (laterally transferred regions) devoid of coding, GC
skew - to predict locations of replication origins and termini in prokaryotes
(iii) k=3: codon usage – for identifying protein-coding regions, frequency of stop codons to
identify open reading frames, Compute codon adaptation index (CAI) that provides an
indication of gene expression level - it compares the distribution of codons actually used in a
particular protein with the preferred codons for highly expressed genes,

8. In BLAST database search algorithm, the match/mismatch ratio for comparing nucleotide
sequences is chosen to be large for highly conserved sequences, while it is small for
divergent sequences. Give reasons, why?
[2]
Soln: When you are looking for very conserved sequences, you expect to see very low %age of
mismatch, hence you want to penalize it heavily in this case. While in the case of divergent
sequences, we expect to see more positions with mismatch, and hence we would want to penalize
it not so heavily in this case.

9. Which of the two scoring matrices, BLOSUM80 and BLOSUM45 would you use for
comparing closely related sequences and evolutionarily distant sequences? [1]
Soln: BLOSUM80 - closely related sequences and BLOSUM45 - evolutionarily distant
sequences.
10. In the four-nucleotide DNA code, the 20 amino acids are encoded in codons of length three.
Suppose Martians have same 20 amino acids but have a five-nucleotide code (A, C, G, T,
Z). What is the minimum codon length required to encode Martian proteins? Justify your
answer.
[2]
Soln: 5^2 = 25, more than 20. So the minimum codon length required in this case is 2.
11. Gene finding is easier in prokaryotes compared to eukaryotes. Give two reasons to support
this statement. [2]
Soln: (i) contiguous coding region in case of prokaryotic gene compared to eukaryotic genes.
(ii) high gene density in prokaryotes.
(iii) exons are not multiple of 3, leading to frame-shift errors.
(iv) stop codons may be present in the intronic regions.

12. You are given a set of ESTs belonging to the same gene locus. Suggest a simple approach to
identify the correct order of the ESTs to build the mRNA sequence? [2]
Soln: By concatenating all the ESTs and obtaining a self-dot plot, the correct order of ESTs can
be identified.
Or, using a program such as polydot in EMBOSS, to obtain an all-against-all dotplots of a
set of sequences.

13. Suppose you have extracted some mRNA and do not know what gene it corresponds to, or
the genome of the organism is not available. How would you design primers for this
sequence?
[2]
Soln: First obtaining cDNA, then inserting it into a bacterial plasmid of known sequence
(cloning) using restriction enzymes. Using primers for the plasmid vector DNA, sequence
the unknown DNA and then create primers for the sequence of interest.

14. A template strand of DNA is 3′ TACCGATTGCA 5′. Which of the following DNA strand
that is created from this template during replication? [1]
A. 5′ TGCAATGCCTA 3′
B. 5′ TAGGCATTGCA 3′
C. 5′ AUGGCUAACGU 3′
D. 5′ ATGGCTAACGT 3′
Soln: D

15. Presence of which of the following cell structures indicates that it is a eukaryotic cell: [1]
A. DNA
B. Nucleus
C. Cytoplasm
D. Ribosomes
Soln: B

16. Which of the following two alignments is preferable? Give reason. [2]
A.
ACAAT
 A -A – T
B.
ACAAT
A - - AT
Soln: B. because in alignment two indels are represented as a single mutational event.

17. Unequal usage of codons in the coding regions appears to be a universal feature of the genomes
across the phylogenetic spectra. What is the likely cause for this bias? [2]
Soln: (i) the uneven usage of the amino acids in the existing proteins and (ii) the uneven usage of
synonymous codons.

18. In the Jukes Cantor model, a CT transition is more likely than a CG transversion. [1]
Soln: False

19. Fourier transform of a coding sequence would result in a peak at 3. [1]

Soln: False. It's at p = 1/3.
PART – B: Answer in the additional answer sheets. (Marks: 58)

1. Under what situations would you use the following variants of dynamic programming
algorithm (give examples): (i) global alignment, (ii) local alignment, (iii) suboptimal
matches, (iv) overlap matches, (v) Linear space alignment algorithm.
[5]
Soln: (i) global alignment – when the sequences are closely related and we can expect an end-to-
end alignment.
(ii) local alignment – when the sequences are distantly related, during database search to identify
homologs, identify conserved regions/motifs, identify common domains shared by two
sequences, etc.
(iii) suboptimal matches – to identify multiple-copies of a domain/motif, more than one
conserved region between a pair of sequences.
(iv) overlap matches – in sequence assembly of genomic sequences, in gene assembly, in EST
clustering, using ESTs for identifying genes in a genome.
(v) Linear space alignment algorithm – while comparing large genomic sequences, to reduce the
space complexity.

2. (a) Give the (i) boundary conditions and (ii) recursive relations in the dynamic program
variant for genomic sequence assembly.
(b) What changes in the above conditions will lead to global alignment?
(c) Why the conditions for obtaining global alignment not suitable for sequence assembly?
(d) Which is more important in sequence alignment: the choice of alignment (global or local)
or the scoring scheme? Give your reasons. [6:2,1,2,1]
Soln: (a) F(0,0) = 0, F(i,0) = 0 = F(0,j), F(i,j) = max{F(i,j)+sij, F(i-1,j)-d, F(i,j-1)-d}
(b) change in the boundary conditions: F(i,0) = -id, F(0,j) = -jd
(c) We do not want to penalize the over-hanging ends.
(d) The most reasonable approach is to use the program based on the appropriate algorithm for
the analysis at hand, and then to choose the scoring system carefully. Small changes in the
scoring system can abruptly change an alignment from a local to a global one.

3. (a) What is the major difference between the approach of FASTA and BLAST in identifying
the seed for extending matches?
(b) How is the significance of an alignment evaluated in the BLAST program?
(c) When would you use PSI-BLAST ? [6:2,2,2]
Soln: (a) BLAST is based on the motivation that - Good alignments should contain many close
matches while FASTA is based on the assumption that - Good alignments should contain
many exact matches. Thus, in BLAST each word in the database sequence, a list of
“neighbourhood” words, scoring above a threshold T are considered as a ‘match’.
(b) If the score of the alignment observed is no better than might be expected from a random
permutation of the sequence, then it is likely to have arisen by chance. For computing the
statistics one needs random sequences. Align these random sequences using the Smith-
Waterman algorithm and compute the score for the optimal alignment. Scores of optimal
alignment follow an extreme value distribution. Measure the mean and standard deviation of
the scores of the alignments of randomized sequences. If the randomized sequences score as
well as the original one, the alignment is unlikely to be significant.
(d) Position Specific Iterated (PSI) BLAST has been designed for finding remote homologues.
When sequence identities dip below 25%, BLAST normally fails to recognize the similarity.
PSI-BLAST is able to work in the range 15%-25% sequence identity levels by using a
scoring matrix tailor-made to find sequences similar to the query from the initial BLAST
search. The process is iterated until you get no more new hits and the algorithm has
converged.

4. (a) There are two major problems with the progressive approach for multiple sequence
alignment: (i) the local minimum problem, and (ii) the choice of alignment parameters. How
can these two problems be addressed?
(b) In ClustalW, what is the justification of retaining gaps introduced in the earlier
alignments?
(c) Why are sequences weighted in ClustalW? [8: 4,2,2]
Soln: (a) (i) The local minimum problem - dependence on the initial pairwise alignment. The
more distantly related these sequences are, the more errors will get propagated through the
alignment. To correct for errors introduced by initial alignment, iterative methods may be
used. Iterative methods attempt to avoid this by repeatedly aligning subgroups of sequences.
(ii) The choice of alignment parameters: when adding more distantly related sequences to the
msa, we are dealing with different evolutionary distances so that the substitution matrix and
the gap penalties must be different at every stage of merging the clusters.
(b) The positions of the gaps that were introduced during the early alignments of the closely
related sequences are not changed as new sequences are added. This is justified because the
placement of gaps in alignments between closely related sequences is much more accurate
than between distantly related ones.
(c) Sequences are weighted to correct for unequal sampling across all evolutionary distances in
the data set. This down-weights sequences that are very similar to other sequences in the data set
and up-weights the most divergent ones.

5. Write short notes on (i) discuss Hirschberg’s algorithm for aligning two sequences, and (ii)
Importance of bootstrapping in phylogenetic analysis. [6]
Soln: (i) Hirschberg’s alignment algorithm uses the principle of divide and conquer for aligning
two sequences.

Steps: Let the two sequences be of size m and n respectively.

Let u = m/2 (integer part)
Identify a v such that the cell (u, v) is on the optimal alignment, i.e., v is the column where the
alignment crosses the i = u row of the matrix
- this splits the dynamic programming problem into two subproblems: from (0, 0) to (u, v), &
from (u, v) to (m, n)
- the full alignment will be the concatenation of the optimal alignments for these two separate
submatrices
This is done recursively, by successively halving each region, until sequences of zero length are
being aligned, or alternatively, sequences are short enough and standard O(nm) alignment and
traceback method can be used.
But how do we find v?
By combining the results of “forward” and “backward” dynamic programming passes at row u,
for each point along the middle row, i.e.,
- the optimal score from (0, 0) to each point in row u and the optimal score from that point to
(m,n)
Adding these numbers gives the optimal score over all paths from (0, 0) to (m, n) that pass
through all the points in row u. A sweep along the middle row, checking these sums, determines
a point (m/2, v) where an optimal path crosses the middle row. This is then done recursively.
(ii) Bootstrapping:
Once a tree has been chosen by some method, it would be desirable to quantify how confident
one is of it. This is given by the statistical technique - bootstrapping.
In this procedure, the true data sequences are used to create a set of new pseudo-replicate
sequences of the same length. Bases at a particular site in the new sequences are chosen to be the
bases appearing in a randomly chosen site in the original sequences. Considering pseudo-
replicates is similar to randomizing a given sequence; to construct a random sequence, we may
call randomly any character at each position to construct a new random sequence, here we call
any random column to construct a new MSA and a new tree corresponding to this MSA.
In bootstrap analysis:
• All sites are considered independent as if freely available in a ‘hat’ to pick
• Available sites are picked up randomly to reconstruct a new alignment of the original size
and a new phylogeny
Process is repeated many times to ascertain the strength of clustering. New “alignment” may
contain several sites multiple times while some other sites may be absent - sampling with
replacement - as a result same column may occur a number of times in the new ‘random’
alignment.
Phylogenies are compared to calculate values [Bootstrap value] that signify the number of times
a given branch/cluster occurred in the Multiple bootstrap trees
Higher the value - higher the confidence of phylogenetic inference
In general values < 50% provide very poor support

6. (a) Define an open reading frame. Discuss issues with using open reading frame (ORF)
approach for gene identification.
(b) In computational gene prediction, give the advantage and limitation of homology-based
approach.
(c) Briefly describe a model-independent approach for finding genes. [10:3,3,4]
Soln: (a) A long sequence between two stop codons devoid of stop codons in-between is called
an ORF.
Addition or deletion of one or two bases will cause all the codons scanned to be different -
sensitive to frame shift errors
 Fails to identify very small coding regions
 In general, the largest ORF is the one that codes for proteins – need not be always true.
 Fails to identify the occurrence of overlapping long ORFs on opposite DNA strands (genes
and ‘shadow genes’)
 Overlapping genes on the same strand observed in bacterial genomes – an overlap of 2-3
bases in an operon.
In the case of eukaryotes,
 due to the existence of interweaving exons and introns – stop codons may exist in intronic
regions making it difficult to identify the correct ORF
 a gene region may encode many proteins – due to alternative splicing/alternative translation
initiation
 Exon length need not be multiple of three – resulting in frameshift between exons
 Gene may be intron-less (single-exon genes)
 Relatively low gene density - only 2 - 3% of the human genome codes for proteins
(b) The genes whose homologs are present in the database can be readily identified by
homology-based database search approach – with the database sizes increasing exponentially,
this is now routine first step in gene prediction and nearly ~50% of genes can be thus identified.
Limitation - only homologs of known genes can be identified, novel genes, or genes specific to
an organism cannot be identified by this approach, and apparent homology can be sometimes
misleading.
(c) Refer to ppt

7. (a) The recognition site for Sau3A I is GATC and is contained in the recognition site of
BamH I, GGATCC. Will the two REs give the same results? If not, which one will give
larger number of fragments?
(b) The enzymes BamH I and Bal II recognise different sequences but leave the same sticky
ends:
BamH I: ----------G|G A T C C ------

Bal II: ----------A|G A T C T ------

Will the two enzymes result in the same number of fragments in a random DNA sequence?
Give reasons.
(c) Construct a restriction map of a linear fragment of DNA, using the following data. Give
the fragment lengths and their positions on a linear scale. [6:2,2,2]
Sizes of
DNA
Fragments (bp)
uncut DNA 10,000
DNA cut with EcoRI 8000, 2000
DNA cut with BamHI 5000, 5000
DNA cut with EcoRI + BamHI 5000, 3000, 2000
Soln: (a) No. The cut made by BamHI is between GG and CC, while by Sau3A I is between GA
and TC. The one with smaller restriction site will make larger number of fragments, since a
pattern of four will occur more often than a pattern of size six.
(b) Yes, because the size of the pattern is the same, and all 6-mers will occur with the same
frequency in a random DNA sequence.
(c) F1: L=2000, Pos: 1 – 2000,
 F2: L=3000, Pos: 2001 – 5000
F3: L=5000, Pos: 50001 – 10,000

The evolutionary distance given by Kimura 2-parameter model accurate as it considers different
substitution rates for transitions and transversions, while Jukes-Cantor model assumes them to
occur at the same rate.

8. (a) Use the distance table below and apply the UPGMA method to compute distances and
build a tree.
S1 S2 S3 S4
S1 .45 .27 .53
S2 .40 .50
S3 .62

(b) Give all possible unrooted trees for any four taxa A, B, C, D. [6:3,3]
Soln: Refer to ppt.

9. Consider the trees in the figure below: [5]

(a) Which of them are the same, as rooted metric trees?
(b) Which of them are the same, as unrooted metric trees?
(c) Which of them are the same, as rooted topological trees?
(d) Which of them are the same, as unrooted topological trees?
(e) For which trees does a molecular clock appear to be operating?
Soln: (a) Same rooted metric trees: T2, T3 [Hint: compute the distance matrix between all taxa,
trees exhibiting same distances between taxa are similar rooted trees, if there rooted topology
exhibits same distances between internal nodes also]
(b) Same unrooted metric trees: T2, T3, T5 [Hint: compute the distance matrix between all taxa,
trees exhibiting same distances between taxa are similar rooted trees, if there topology is also
the same]
(c) Same rooted topological trees: T2, T3, T5 and T1, T6 [Hint: same rooted branching pattern]
(d) Same unrooted topological trees: all [Hint: same branching pattern, exhibiting same
neighbours]
(e) Trees for which molecular clock appear to be operating: T4 and T6 [Hint: compute the
distance of each taxa from the root, if all the taxa are placed at equidistant from the root,
molecular clock is operating]