advanced.qmd

---
title: "Advanced homology modeling"
author: "Modesto Redrejo Rodríguez"
date: "`r Sys.Date()`"
toc: true 
toc_float: true
format:
  html:
    theme: simplex
    toc: true
    toc-location: right
    toc-depth: 4
    number-sections: false
    code-overflow: wrap
    link-external-icon: true
    link-external-newwindow: true
bibliography: references.bib
editor_options: 
  markdown: 
    wrap: 80
---

Although we do not intend to compile the evolution of modeling methods, I
briefly outline below the origin and transformation of advanced protocols that
outperform single-template homology modeling during the last three decades. This
step-wise evolution of modeling methods is the origin of the revolution of
Alphafold and related protocols, which we will discuss in the next
[section](ai.html).

# From homology modeling to threading

## 1D features prediction

By definition 1D features are protein features that can be decoded directly from
the protein primary structure and represented as values (categories, %, ...)
associated to individual residues in the sequence. For instance, we can assign a
secondary structure state (symbol or probability) to each residue. Many
structure prediction methods implement or call third parties methods to predict
secondary structure and other 1D features, as important additional information
during modeling process.

You can find links to several 1D features prediction tools in the [Modeling
Resources](links.html) section.

### Secondary structure prediction

Regarding experimentally-based structures, secondary structures can be
automatically assigned using **DSSP** (*Define Secondary Structure of Proteins*)
algorithm, originally written in 1983 and updated several times throughout the
years, being the last version from 2021 (available on
[GitHub](https://github.com/PDB-REDO/dssp)). This algorithm classifies each
residue considering its geometry and H-bonds prediction by comparison with
pre-existing patterns in DSSP database. Remarkably, [DSSP does not predict]{.ul}
secondary structures, it just extracts this information from the 3D coordinates.

Prediction of secondary structure (SS) from protein sequences is based in the
hypothesis that Segments of consecutive residues have preferences for certain
secondary structure states. Similar to other methods in bioinformatics,
including protein modeling, approaches to SS prediction evolved during the last
50 years (see Table 1).

First generation methods rely on statistics approaches and prediction depends on
assigning a set of prediction values to a residue and then applying a simple
algorithm to those numbers. I.e. apply a probability score based on single
amino acid propensity. In the 1990's, new algorithms included the information of
the flanking residues (3-50 nearby amino acids) in the so-called Nearest
Neighbor (N-N) Methods. These methods increased the accuracy in many cases but
still had strong limitations, as they only considered three possible states
(helix, strand or turn). Moreover, as you know from the secondary structure
practice, β-strands predictions are more difficult and did not improve much
thanks to N-N methods. Additionally, predicted helices and strands were usually
too short.

By the end of 1990 decade, new methods boosted the accuracy to values near to
80% and even 90%. These methods included two innovations, one conceptual and one
methodological. The conceptual innovation was the inclusion of evolutionary
information in the predictions, by considering the information of multiple
sequence alignments or profiles. If a residue or a type of residue is
evolutionary conserved, it is likely that it is important to define SS
stretches. The second innovation was the use of neural networks (see
[below](#sec-NN)) in which multiple layers of sequence-to-structure predictions
were compared with a independently trained networks (see PHD paper by Burkhard
Rost [here](https://www.rostlab.org/papers/1996_phd/paper.html)).

In the last years, most commoly used methods are meta-servers that compare
several algorithms, mostly based o neural-networks, like
[JPred](http://www.compbio.dundee.ac.uk/jpred/) or
[SYMPRED](https://www.ibi.vu.nl/programs/sympredwww/), among others.

+------------------+-------------------------------------+--------------------+
| **Type**         | **Method**                          | **Accuracy**       |
+------------------+-------------------------------------+--------------------+
| Statistics       | [Chow &                             | 57%                |
|                  | Fassman](http:%20%20%20%20%20       |                    |
|                  | //www.biogem.org/tool/chou-fasman/) |                    |
|                  | (1974-)                             |                    |
+------------------+-------------------------------------+--------------------+
|                  | [G                                  | 63-73.5%\          |
|                  | OR](h%20ttp%20s:/%20/np%20sa-%20pra | (Version V)        |
|                  | bi.ibcp.fr/cgi-bin/n%20%20%20%20%20 |                    |
|                  | psa_automat.pl?page=npsa_gor4.html) |                    |
|                  | (1978-)                             |                    |
+------------------+-------------------------------------+--------------------+
| Nearest Neighbor | PREDATOR (1996)                     | 75%                |
| (N-N) methods    |                                     |                    |
+------------------+-------------------------------------+--------------------+
|                  | NNSSP (1995)                        | 72%                |
+------------------+-------------------------------------+--------------------+
| N-N neural       | [APSSP](h%20%20%20%20%20            | Up to 86%          |
| network          | ttp://imtech.res.in/raghava/apssp/) |                    |
+------------------+-------------------------------------+--------------------+
|                  | [PsiPRED](h%20%20%20%20%20          | 75.7% (1999)       |
|                  | ttp://bioinf.cs.ucl.ac.uk/psipred/) |                    |
|                  | (1999-)                             | 84% (2019)         |
+------------------+-------------------------------------+--------------------+
|                  | [PHD](h                             | 74%                |
|                  | ttps:%20//n%20psa%20-pr%20abi%20.ib |                    |
|                  | cp.fr/cgi-bin/npsa_a%20%20%20%20%20 |                    |
|                  | utomat.pl?page=/NPSA/npsa_phd.html) |                    |
|                  | (1997)                              |                    |
+------------------+-------------------------------------+--------------------+
| HMM              | [SAM](http%20%20%20%20%20           | 76%                |
|                  | s://compbio.soe.ucsc.edu/HMM-apps/) |                    |
+------------------+-------------------------------------+--------------------+
| **META-Servers** | [                                   |                  |
|                  | **Jpred4**](http://w%20%20%20%20%20 |                    |
|                  | ww.compbio.dundee.ac.uk/www-jpred/) |                    |
+------------------+-------------------------------------+--------------------+
|                  | GeneSilico (Discontinued)           |                    |
+------------------+-------------------------------------+--------------------+
|                  | [SYMPRED](https://%20%20%20%20%20   |                    |
|                  | www.ibi.vu.nl/programs/sympredwww/) |                    |
+------------------+-------------------------------------+--------------------+

: Evolution of secondary structure prediction methods

### Structural disorder and solvent accessibility

The expression *disorder* denote protein stretches that cannot be assigned to
any SS. They are usually dynamic/flexible, thus with high B-factor or even
missing in crystal structures. These fragments show a low complexity and they
are usually rich in polar residues, whereas aromatic residues are rarely found
in disordered regions. These motifs are usually at the ends of proteins or
domain boundaries (as linkers). Additionally, they are frequently related to
specific functionalities, such in the case of proteolytic targets or
protein-protein interactions (PPI). More rarely, large disordered domains can be
conserved in protein families and associated with relevant functions, as in the
case of some transcription factors, transcription regulators, kinases...

There are many methods and servers to predict disordered regions. You can see a
list in the Wikipedia
[here](https://en.wikipedia.org/wiki/List_of_disorder_prediction_software) or in
the review by @atkins2015. The best-known server is
[DisProt](https://www.disprot.org/), which uses a large curated database of
intrinsically disordered proteins and regions from the literature, which has
been recently improved to version 9 in 2022, as described in @quaglia2022.

Interestingly, a low plDDT (see [below](ai.html#sec-AF2)) score in Alphafold2
models has been also suggested as a good indicator of protein disorder
[@wilson2022].

![Examples of disordered regions. From [Database of disordered
regions](http://original.disprot.org/index.php).](pics/imageSwap.gif){#fig-disprot
.figure}

*Hydrophobic collapse* is usually referred to as a key step in protein folding.
Hydrophobic residues tend to be buried inside the protein, whereas hydrophilic,
polar amino acids are exposed to the aqueous solvent.

![Hydrophobic collapse as a early step in protein folding. From Feenstra &
Abeln, [Introduction to Structural
Bioinformatics](https://www.studocu.com/row/document/bangladesh-agricultural-university/protein-folding/introduction-to-structural-bioinformatics/11627608).](pics/collapse.png "Hydrophobic collapse"){#fig-collapse
.figure}

Solvent accessibility correlates with residue hydrofobicity (accessibility
methods usually better performance). Therefore, estimation of how likely each
residue is exposed to the solvent or buried inside the protein is useful to
obtain and analyze protein models. Moreover, this information is useful to
predict PPIs as well as ligand binding or functional sites. Most methods only
classify each residue into two groups: Buried, for those with relative
accessibility probability \<16% and Exposed, for accessibility residues \>16%.

Most common recent methods, like [ProtSA](http://webapps.bifi.es/ProtSA/) or
[PROFacc](www.rostab.org), combine evolutionary information with neural networks
to predict accessibility.

### Trans-membrane motifs and membrane topology

Identification of transmembrane motifs is also a key step in protein modeling.
About 25-30% of human proteins contain transmembrane elements, most of them in
alpha helices.

![Different topologies of transmembrane proteins](pics/mb.png){#fig-TM .figure}

The PDBTM ([Protein Data Bank of Transmembrane
Proteins](http://pdbtm.enzim.hu/)) is a comprehensive and up-to-date
transmembrane protein selection. As of September 2022, it contains more than
7600 transmembrane proteins, 92.6% of them with alpha helices TM elements. This
number of TM proteins is relatively low, as compared with more than 160k
structures in PDB, as TM proteins are usually harder to purify and
crystalization conditions are often elusive. Thus, although difficult, accurate
predictions of TM motifs and overall protein topology can be essential to define
protein architecture and identify domains that could be structurally or
functionally studied independently.

![Different topologies of TM helices.](pics/alphaTM.jpg){#fig-topo .figure}

Current state-of-the-art TM prediction protocols show an accuracy of 90% for
definition of TM elements, but only a 80% regarding the protein topology.
However, some authors claim that in some types of proteins, the accuracy is not
over 70%, due to the small datasets of TM proteins. Most recent methods, based
in deep-learning seem to have increased the accuracy to values near 90% for
several groups of proteins [@hallgren].

### Subcellular localization tags and post-translational modification sites

Many cellular functions are compartmentalized in the nucleus, mitochondria,
endoplasmatic reticulum (ER), or other organules. Therefore, many proteins
should be located in those compartments. That is achieved by the presence of
some labels, in form of short peptidic sequences that regulate traffic and
compartmentalization of proteins within the cells. Typically, N-terminal signals
direct proteins to the mitochondrial matrix, ER, or peroxisomes, whereas nucleus
traffic is regulated by nuclear localization signals (NLS) and nuclear export
signals (NES).

Similarly, post-translational modifications very often occur in conserved motifs
that contain target residues for phosphorylation or ubiquitination, among other
modifications.

These short motifs are difficult to predict, as datasets of validated signals
are small. The use of consensus sequences allowed predictions, although in many
cases with a high level of uncertainty.

## Threading or Fold-recognition methods {#sec-threading}

As mentioned earlier, the introduction of HMM-based profiles during the first
decade of this century led to a great improvement in template detection and
protein modeling in the twilight zone, i.e., proteins with only distant homologs
(\<25-30% identity) in databases. In order to exploit the power of HMM searches,
those methods naturally evolved into iterative [threading]{.ul} methods, based
on multitemplate model construction, implemented in
[I-TASSER](https://zhanggroup.org/I-TASSER/) [@roy2010],
[Phyre2](http://www.sbg.bio.ic.ac.uk/phyre2/) [@kelley2015], and
[RosettaCM](https://new.rosettacommons.org/demos/latest/tutorials/rosetta_cm/rosetta_cm_tutorial)
[@song2013], among others. These methods are usually referred to as
**Threading** or **Fold-recognition** methods. Note that the classification of
modeling methods is often blurry. The current version of SwissModel and the use
of HHPred+Modeller already rely on HMM profiles for template identification and
alignment; being thus strictly also fold-recognition methods.

Both terms can be often used interchangeably, although some authors see
**Fold-Recognition** as any technique that uses structural information in
addition to sequence information to identify remote homologies, while
**Threading** would refer to a more complex process of modeling including remote
homologies and also the modeling of pairwise amino acid interactions in the
structure. Therefore, although we have already used HHPred along with the use of
HHPred to identify templates for subsequent modeling, could be indeed considered
threading.

![The idea behind fold-recognition is that instead of comparing sequences, we
intend to compare structures. In the Frozen approximation (left), one residue is
aligned with the template structure and then we evaluate the probability of the
nearby residues in the query sequence to be in the same position than the
equivalent in the template. On the other hand, Defrost methods use profiles to
generate improved alignments that allow better starting points to the energy
calculations during the iterative modeling steps. From
@kelley2009a.](pics/fr.png){#fig-frost .figure}

The Iterative Threading ASSembly Refinement
([I-TASSER](http://zhanglab.ccmb.med.umich.edu/I-TASSER)) from [Yang Zhang
lab](https://zhanggroup.org/) is one of the most widely used threading methods
and servers. This method was was ranked as the No 1 server for protein structure
prediction in the community-wide [CASP7](https://zhanggroup.org/casp7/21.html),
[CASP8](https://predictioncenter.org/casp8/groups_analysis.cgi?target_type=0&gr_type=server),
[CASP9](https://predictioncenter.org/casp9/CD/data/html/groups.2.html),
[CASP10](http://predictioncenter.org/casp10/groups_analysis.cgi?type=server),
[CASP11](http://www.predictioncenter.org/casp11/zscores_final.cgi?gr_type=server_only),
[CASP12](http://www.predictioncenter.org/casp12/zscores_final.cgi?&gr_type=server_only),
[CASP13](http://www.predictioncenter.org/casp13/zscores_final.cgi?model_type=best&gr_type=server_only),
and
[CASP14](http://www.predictioncenter.org/casp14/zscores_final.cgi?gr_type=server_only)
experiments. I-TASSER first generates three-dimensional (3D) atomic models from
multiple threading alignments and iterative structural assembly simulations that
are iteratively selected and improved. The quality of the template alignments
(and therefore the difficulty of modeling the targets) is judged based on the
statistical significance of the best threading alignment, i.e., the *Z*-score,
which is defined as the energy score in standard deviation units relative to the
statistical mean of all alignments.

![Flowchart of I-TASSER protein structure modeling. From
@rigden2017a.](images/paste-29011FF5.png){#fig-tasser .figure}

First, I-TASSER uses Psi-BLAST against curated databases to select sequence
homologs and generate a sequence profile. That profile is used to predict the
secondary structure and generate multiple fragmented models using several
programs. The top template hits from each threading program are then selected
for the following steps. In the second stage, continuous fragments in threading
alignments are excised from the template structures and are used to assemble
structural conformations of the sections that aligned well, with the unaligned
regions (mainly loops/tails) built by *ab initio* modeling. The fragment
assembly is performed using a modified replica-exchange Monte Carlo random
simulation technique, which implements several replica simulations in parallel
using different conditions that are periodically exchanged. Those simulations
consider multiple parameters, including model statistics (stereochemical
outliers, H-bond, hydrophobicity...), spatial restraints and amino acid pairwise
contact predictions ([see below](#maps)). In each step, output models are
clustered to select the representative ones for the next stage. A final
refinement step includes rotamers modeling and filtering out steric clashes.

One interesting thing about I-TASSER is that it is integrated within a server
with many other applications, including some of the tools that I-TASSER uses and
other advanced methods based on I-TASSER, like I-TASSER-MTD for large,
multidomain proteins or C-I-TASSER that implements a deep learning step, similar
to Alphafold2 (see [next section](ai.html)).

![RosettaCM Protocol. (A) Flowchart of the RosettaCM protocol. (B--D) RosettaCM
conformational sampling. From
@song2013.](pics/rosettaCM.jpg "RosettaCM Protocol"){#fig-rosettacm .figure
width="664"}

**RosettaCM** is an advanced homology modeling or threading algorithm by the
[Baker lab](https://www.bakerlab.org/), implemented in
[Rosetta](https://www.rosettacommons.org/software) software and the
[Robetta](https://robetta.bakerlab.org/) webserver. RosettaCM assemblies the
model by recombining aligned segments in a set of selected templates and close
the model by a minimization of the junctions between segments by fragments
torsion and iterative optimization steps that include Monte Carlo sampling.
Finally, an *all-atom* refinement towards a minimum of free energy [@song2013].

# From contact maps to pairwise high-res feature maps {#maps}

![Contact-based map of representative proteins. The map represents a matrix of
amino acid positions in the protein sequences (on both, the X and Y axis); with
contacts indicated as blue dots. When a number of consecutive residues in the
sequence interact the dots form diagonal stretches. Maps obtained at
<http://cmweb.enzim.hu/>](pics/contact.png "Contact-based map"){#fig-contact
.figure}

During the last decade, the introduction of **residue-residue contact or
distance maps** prediction based on sequence co-evolution and *deep learning*
started a revolution in the field that crystallize with the arrival of
Alphafold2 and RoseTTAfold as major breakthroughs with great repercussions in
diverse fields. As shown in the [Figure 9](#fig-contact), residue-residue
contact maps can be obtained from structures as a matrix that show close
residues and show a pattern that clearly show differences between motifs and
secondary structure stretches.

![Schematic of how co-evolution methods extract information about protein
structure from a multiple sequence alignment (MSA). From
@bittrich2019](pics/contact_evol.png){#fig-coevol .figure}

An accurate information of protein's residue--residue contacts is sufficient to
elucidate the fold of a protein [@olmea1997]; however implementation of these
maps in protein modeling is not trivial, as predicting that map is not always
easy. The introduction of **direct-coupling analysis (DCA)**, i.e., extract the
residue coevolution from MSAs ([Figure 4](#fig-coevol)) improved contact maps
and allowed their implementation for protein folding in several methods, like
PSICOV [@jones2012] or Gremlin [@kamisetty2013], among others. However, it
should be noted for proteins without many sequence homologs, the predicted
contacts were of low quality and insufficient for accurate contact-assisted
protein modeling.

![Illustration of column pair and precision sub-matrix grouping for advanced
prediction of contact maps. In the example, Columns 5 and 14 in the first family
are aligned to columns 5 and 11 in the second family, respectively, so column
pair (5,14) in the first family and the pair (5,11) in the second family are
assigned to the same group. Accordingly, the two precision sub-matrices will be
assigned to the same group. From @ma2015.](pics/contact2.png){#fig-families
.figure}

### Implementation of several layers of information processed by neural network and deep learning methods {#sec-NN .section}

Deep learning is a sub-field of machine learning which is based on artificial
neural networks (NN). Neural networks were introduced actually in the late 40's
and 50's, but they reappeared in the 2000's thanks to the increase in
computational capacities and, more recently, the use of
[GPUs](https://en.wikipedia.org/wiki/Graphics_processing_unit). Briefly, a NN
uses multiple interconnected layers to transform multiple inputs (MSAs,
high-resolution contact-based maps...) into compound features that can be used
to predict a complex output, like a 3D protein structure. As their name
indicates, NNs attempt to simulate the behavior of the human brain that
processes large amounts of data and can be trained to "learn" from that data.
Deep learning is based on the use of multiple layer-NN to optimize and refine
for accuracy.

The next level of complexity in contact maps is their application to distantly
related proteins by comparing sets of DCA from different protein families, a
method sometimes referred as joint evolutionary coupling analysis ([Figure
10](#fig-families)). This kind of analysis entails processing a huge amount of
information, which increases the computational resources requirements. Thus, the
use of trained neural networks and state-of-the-art deep-learning methods
boosted the capacities of protein modeling.

In this context, the introduction of supervised machine learning methods that
predict contacts, outperforms DCA methods by the use of multilayer neural
networks [@jones2015; @ma2015; @wang2017a; @yang2020]. These methods implemented
the use of the so-called high resolution contact maps, which contains enriched
information with not only contacts but also distances, and angles, represented
in a heatmap-like probability scale.

![Example of high-resolution contact maps of 6MSP. From
@yang2020](pics/high_res_maps.png "High-resolution contact maps"){#fig-highres
.figure}