Open source SMFS

1. Proteins
2. Atomic force microscopy
3. Single molecule force spectroscopy
4. Experiment control
5. Cantilever calibration
6. Modeling: Monte Carlo unfolding simulations
7. Application: Unfolding in salty buffers
8. Conclusions and future work

Proteins: What are they?

Proteins: Where are they?

Takamori, Holt, Stenius, et al., 2006

Proteins: Titin

Adapted from Wikipedia

Proteins: Titin's I27

U. Illinois Biophysics Group

Proteins: I27

PDB structure 1TIT

Proteins: What's the problem?

MHHHHHHSSLIEV EKPLYGVEVFVGE TAHFEIELSEPDV HGQWKLKGQPLTA SPDCEIIEDGKKH ILILHNCQLGMTG EVSFQAANAKSAA NLKVKEL	→
Pirchi, Ziv, Riven, et al., 2011

Open source SMFS

1. Proteins
2. Atomic force microscopy
3. Single molecule force spectroscopy
4. Experiment control
5. Cantilever calibration
6. Modeling: Monte Carlo unfolding simulations
7. Application: Unfolding in salty buffers
8. Conclusions and future work

Atomic force microscopy

AFM: Cantilever geometry

Olympus TR800PSA , images from Asylum Research
We use the thinner TR400PSA

AFM: Laser deflection

Charles Roduit

AFM: Piezo positioning

The piezoelectric effect

Lead zirconium titanate (PZT) from Wikipedia

AFM: Tubular piezos

Open source SMFS

1. Proteins
2. Atomic force microscopy
3. Single molecule force spectroscopy
4. Experiment control
5. Cantilever calibration
6. Modeling: Monte Carlo unfolding simulations
7. Application: Unfolding in salty buffers
8. Conclusions and future work

Single molecule force spectroscopy

SMFS: Sawtooth curve

SMFS: What's going on?

Carrion-Vazquez, et al., 2000; adapted from Baljon and Robbins, 1996

SMFS: Unfolding one domain

Lu and Schulten, 2000

Open source SMFS

1. Proteins
2. Atomic force microscopy
3. Single molecule force spectroscopy
4. Experiment control
5. Cantilever calibration
6. Modeling: Monte Carlo unfolding simulations
7. Application: Unfolding in salty buffers
8. Conclusions and future work

Experiment control

Control: Quick-and-dirty

Control: My modular stack

Open source: Existing layers

Linux	GNU	Gentoo	Python	SciPy
Comedi	matplotlib	pymodbus	Cython	NumPy
h5py	…

Open source: Teamwork

Mlot, Tovey, and Hu, 2011

Control: Example code

class Unfolder (object):
    …
    def run(self):
        """Approach-bind-unfold-save[-plot] cycle.
        """
        ret = {}
        ret['timestamp'] = _email_utils.formatdate(localtime=True)
        ret['temperature'] = self.afm.get_temperature()
        ret['approach'] = self._approach()
        self._bind()
        ret['unfold'] = self._unfold()
        self._save(**ret)
        if _package_config['matplotlib']:
            self._plot(**ret)
        return ret
						

Archival: HDF5 and h5config

GROUP "/"
   GROUP "approach"
      …
   GROUP "config"
      GROUP "afm"
         …
      GROUP "approach"
         …
      DATASET "bind time"
      …
      GROUP "unfold"
         …
         DATASET "velocity"
   GROUP "environment"
      DATASET "temperature"
      DATASET "timestamp"
      …
   GROUP "unfold"
      DATASET "deflection"
      DATASET "frequency"
      DATASET "z"

Archival: Version control

commit 32bfbf98d79c73eba50b77d0917df100e0e33bcf
Author: W. Trevor King <wking@tremily.us>
Date:   Fri Jan 18 22:54:49 2013 -0500

    afm: Optionally return stepper_approach data with `record_data`

    Sometimes these approach curves are pretty funky, so I'll start
    recording them by default in calibcant-calibrate.py.

diff --git a/pyafm/afm.py b/pyafm/afm.py
index 60741c6..e76b118 100644
--- a/pyafm/afm.py
+++ b/pyafm/afm.py
@@ -460,10 +460,11 @@ class AFM (object):
         _LOG.warn(e)
         raise e

-    def stepper_approach(self, target_deflection):
+    def stepper_approach(self, target_deflection, record_data=None):
…

Open source SMFS

1. Proteins
2. Atomic force microscopy
3. Single molecule force spectroscopy
4. Experiment control
5. Cantilever calibration
6. Modeling: Monte Carlo unfolding simulations
7. Application: Unfolding in salty buffers
8. Conclusions and future work

Cantilever calibration

Calibration: Geometry

Olympus TR800PSA , images from Asylum Research
We use the thinner TR400PSA

Calibration: Equipartition

The average spring energy is $$\frac{1}{2}\kappa ⟨x_c^2⟩=\frac{1}{2}{k}_{B}T,$$ where $k_B$ is Boltzmann's constant and $T$ is the temperature.

Calibration: Vibration

Calibration: Photodiode calibration

Calibration: Results

Calibration: Stability

Calibration: Inconsistency

Florin, Rief, Lehmann, et al., 1995

Open source SMFS

1. Proteins
2. Atomic force microscopy
3. Single molecule force spectroscopy
4. Experiment control
5. Cantilever calibration
6. Modeling: Monte Carlo unfolding simulations
7. Application: Unfolding in salty buffers
8. Conclusions and future work

Monte Carlo unfolding simulations

Hooke: Experimental histograms

Sandal, Benedetti, et al., 2009

Sawsim: State model

My simulation framework.

Sawsim: Simulation loop

1. Calculate piezo-induced gap $x_t(t)=vt$
2. Find tension model parameters for each state
3. Distribute per-state stretching ($x_c$, $x_u$, …) to balance the tension
4. Calculate the transition rates between states
5. Roll the dice to determine if transitions take place as you step forward in time

Sawsim: Monte Carlo

Sawsim: Unfolding models

Sawsim: Kramers' model

Sawsim: Tension models

Sawsim: Fitting models

Open source SMFS

1. Proteins
2. Atomic force microscopy
3. Single molecule force spectroscopy
4. Experiment control
5. Cantilever calibration
6. Modeling: Monte Carlo unfolding simulations
7. Application: Unfolding in salty buffers
8. Conclusions and future work

Unfolding in salty buffers

Lu and Schulten, 2000

Salt: Glutamic acid

Salt: Reduced stability in CaCl₂

Salt: Sawsim fits

Ca²⁺ radius ∼1.1 Å, H-bond ∼2 Å.

Open source SMFS

1. Proteins
2. Atomic force microscopy
3. Single molecule force spectroscopy
4. Experiment control
5. Cantilever calibration
6. Modeling: Monte Carlo unfolding simulations
7. Application: Unfolding in salty buffers
8. Conclusions and future work

Conclusions: Salt

• We see the expected destabilizing effect of CaCl₂
• Future work:
  • More contextual data!
    • Pulling speeds
    • Salt concentrations (physiological levels)
    • Salt species
  • Mutated proteins?
    • Is glutamic acid special?

Conclusions: Hardware

Open source AFM automation is possible, even with older hardware originally designed for more manual operation.

Future: Hardware

For automatic control, it would be nice to have… Piezos with capacitive feedback ($10k per axis) Independent 4-segment photodiode readout Control over photodiode positioning Control over laser alignment

Conclusions: Software

New open source software:

• Pycomedi
• Pypiezo
• Stepper
• Pypid
• Pyafm
• Unfold-protein
• Calibcant
• H5config
• Hooke
• Sawsim

Future: Software

Everything works for me, and I expect it will work for others… but no software (except maybe TeX) is without bugs. Testers welcome!

The End

Thanks!