bibliography - stanford universityqs400mf2512/qs400mf2512.pdfuse of different standardsfor em and...

Experimental Reasoning April 11, 1977

A Case Study of the Reasoning in a Genetics Experiment

Heuristic Programming ProjectWorking Paper 77-18

Jerry FeitelsonMark Stefik

Abstract: The laboratory steps for a series of geneticsexperiments are examined in depth and the reasoning and knowledge usedto plan the experiments are characterized. One surprise is the extentto which the planning process seems to be event driven* For thisexperiment, the planning process would not be well characterized as thesearch of a large space for the solution to a fixed experiment.Rather, most planning in these experiments seems to be short term andin response to unexpected results in the laboratory* Considerableknowledge is used in forming new hypotheses in response to theunexpected. Furthermore, much of the geneticist's behavior seems to bedirected toward exploiting serendipity*

Support :

MOLGEN grants NSF MCS76- 11935NSF MCS76- 11649

Training grant NIH STOI GM00295-17

Heuristic ARPA DAHC 15-73-C-0435Programming Project

SUMEX computingfacility

NIH Biotechnology Resources ProgramRR-785

(105) HPP-77-18

y/r

Table of Contents

I

Chapter

I* Introduction . . . m „ m . „ tt %' w # *

ll* Background and Some Definitions " * * * .* „ . .. w w, 2

111. A Description of the Experiment *

'

* «,

■

... » tt „ # „ „. 4

111*1 Overview of the Experiment * * * * „ * » *. * . 4

111*2 The Experimental Steps in Detail *

,*.

* . «. * « , 5

IV* Review of Knowledge Used in This Experiment * * .. .t v , 10

IV. 1 Proposing the Experiment .. » , . * * 8. ,10

IV* 2 Establishing Some Subgoals " . " .. » fe s «, , . 12

IV*3 The First Part of the Experiment *

*,

* w * , „ €l 151V.,4 A Brief Recapitulation * * » , * , g . € 19IV. 5 The Second Half of the Experiment * * * s ,- 9

,

„ 19

IV, 6 Some Experimental Fishing Trips . .. « « » , ,r 24

IV*6*l Transforming B* subtilis with the HybridPlasmids * »

,

» „ ,„

,

s, s s .., 24

IV*6»2 Extending the Colony Hybridization Techniqueto B* subtilis * * % « * . ,» .

.

26

IV*6*3 Back Hybridizing the pFT's to Phi-3-T DNA * .-, » 28

V* Some Thoughts About the Knowledge Used in ThisExperiment *

,*

, m t . , * 30V* 1 General Observations * « . * tl ' «. , „. . . 30

Experimental Reasoning V.4 April 11, 1977

II

V.2 Some Proposed Important Parameters ..****. 33V.3 Rule Classifications . * * * oc

V.4 Concluding Remarks * * . ... v , **..*.* 35

Appendix I

Bibliography 36« *


Chapter I_

Introduction

The MOLGEN group has described a process of experimentPlanning which made much use of hierarchical problem solving to pre-experiraf ntal stePs * T^se experiments have been classified asn^ fl3.^ f YSIS cxPeriments w".h goals explicitly defined by theuser at the beginning of planning. This report explores in depth aseries of^actual genetics experiments performed in Professor JoshuaLederberg s laboratory over a two month period in order to characterizeboth the experiment design process and the knowledge used to guide thedecision making, b tne

In this study, the geneticist is viewed as having a theory inT,^Zri °T beiief? *re heid with different degrees of certainty.His theory describes the limitations of his laboratory and his currentresearch goals.. Given this theory, he must decide what to do next. Hemay plan to do certain experiments and perform certain operations, butunexpected measurements and fortunate observations can steer him indifferent directions.

Judging only from this single case study, relatively littletime is spent evaluating experimental alternatives for a fixedexperiment* The art of successful experimentation involves shrewddecisions of what to do next based on recent observations and the stateof the theory*- A geneticist is an expert at exploiting serendipity andat generating hypotheses to be tested.

Chapter II gives some of the genetics background andterminology for the experiment being studied* Chapter 111 thenelaborates the steps of the experiment in considerable detail. ChapterIV steps through the experiment again and highlights the knowledgeF^nvS rT\ decisionsat critical moments of planning.Finally, Chapter V makes some general observations about theknowledge used and the planning process.

1 See [Martin77] and [Stefik77].2 See [Ehrlich76] and [Ehrlich77]*


& &^£&

Chapter II

Background and Some Definitions

The experiment described involves two kinds of bacteria —Escherichia coli (E. coli) and Bacillus subtilis (B*. subtilis) and thebacteriophage Phi- 3-T* Phi-3-T carries a thymidylate synthetase gene*When Phi-3-T infects Thy" B. subtilis (i.e. lacking in the ability tosynthesize Thymidine), it "transforms it to prototrophy" . This meansthat the Phi- 3-T confers the ability to synthesize thymidine to the B*.subtilis by donating its own genetic information. Thy" B^ subtilisrequires thymidine in the growth medium to survive* We will beconcerned with two plasmids (snail, extrachromosomal, circular DNAmolecules) of Ej. coli in addition to the main bacterial chromosome.These are called pSCIOI and pMB9. Both plasmids carry a gene forresistance to the drug tetracycline* When E* coli lacks this gene, itis sensitive to tetracycline and is said to have the Tc s phenotype.The plasmids have the ability to confer resistance to tetracycline whenthey are taken into E± coli — giving it the Tcr phenotype. They areused as "vectors", that is, carriers for foreign DNA.

B* subtilissu lis The bacterium Bacillus subtilis*E*. coliThy*, Thy"

The bacterium Escherichia coli.Thymidine (T) is a base essential to synthesize

di.

or repair DNA*Thy" bacteria require it in the growth medium.Thy bacteria can synthesize it from other

Tcr , Tcs compounds in the environment.Resistance (or sensitivity) to tetracycline.

Phi-3-T

pSCIOIpMB9pFT

Bacteriophage which can lysogenically convertThy" B* subtilis to Thy+*Plasmid of E*. coli carrying Tcr *Another E*. coli plasmid also carrying Tcr*Hybrid plasmid carrying Phi-3-T gene* Such

ColEl-ampplasmids will hybridize with Phi-3-T cRNA.Another plasmid conferring resistance to theantibiotic ampicillin.

cRNA

EcoRTBamHl

Radioactively-labeled complementary RNA madefrom DNA templates using RNA polymerase.A restriction enzyme from E*. coli.A restriction enzyme from Eg. amvloliouefaclens.

T4 ligase

EM

A ligase enzyme, which can join together DNAsegments which have been cleaved by Ecoß-.Abbreviation for electron microscopy. Tnis

r~m~technique allows viewing many gross featuresof appropriately stained DNA molecules.The phenotype of the E± coli used in this

2

*

3

n-xperimentai Reasoning April 11, 1977

experiment lacks a restriction/modificationsystem for destroying foreign DNA. Thismakes the introduction of plasmids easiertechnically.A region of DNA to which RNA polymerase bindsto initiate transcription.Bacteria with the ability to grow on a minimalmedium* ("Wildtype")Bacteria unable to grow without nutritionalsupplements. This is often due to geneticdefects at defined loci.

promoter

prototroph

auxotroph

I


4

Chapter 111

A Description of the Experiment

For the last two years, Prof* Lederberg 's group has been tryingto transfer a gene from B. subtilis to Ev coli. In this experiment,they modified their goal somewhat by trying instead to transfer a genefrom Phi -3-T, a bacteriophage of B^ subtilis. DNA from thebacteriophage is somewhat easier to handle. It is shorter and easierto obtain in concentrated and purified amounts* The Thy gene in Phi-3-Tis capable of restoring to prototrophy strains of B*_ subtilis which aredeficient in either of the thymidine genes — Thy A or Thy B* Thisexperiment is significant because it supports the hypothesis that genescan be transferred between prokaryotes and expressed (ie. producefunctional products in the new host)* In addition, several interestingobservations made during this experiment have suggested some directionsfor further research*

This chapter describes the experimental steps in detail butgives only scant motivation for the steps or interpretation of results.Chapter IV goes through this experiment again emphasizing theknowledge used in planning the steps and in analyzing the results.

111.1 Overview of the Experiment

I* Extract Phi -3-T DNA from the bacteriophage (and perform varioustests on it).

11. Digest Phi -3-T DNA with the enzyme EcoRT and test for transformingactivity.

Ill* Ligate the Phi-3-T DNA with Ecoß cleaved pSCIOI plasmids tocreate hybrid plasmids.

IV* Transform r" m~ Thy" TcS Ej_ coli to Thy+ Tcr with the clonedplasmids and Isolate the transformants* (This involves reasoningbased on biological function of the genes involved.)

V* Verify that the Thy+ character of the transformants is actuallyconferred by the hybrid plasmid. (Other hypotheses are possibleand must be tested.)

VI. Use heteroduplex analysis to examine the molecular structure of

Experimental Reasoning 111. 1 April 11, 1977

5

the hybrid plasmids to determine which segment contains the Thygene*

VII. Transfer the hybrid plasmids into B. subtilis and test fortransformation .

VIII Since the technique of Heteroduplex Analysis is so time-consuming and not optimal for large bacterial chromosomes, asimple (albeit less specific) method for testing for thepresence of DNA sequences by homology was extended. Inparticular, the in situ colony hybridization technique wastested and verified to work for B. subtilis.

IX. Use the extended colony hybridization technique to examine themolecular structure of the transformed Bj*_ subtilis. A surprisingresult was that the Thy gene was incorporated into the B*_ subtilischromosome but that pSCIOI DNA was not.

X. Reverse the hybridization procedure and to check if the hybridplasmids are homologous to a single band in the Phi-3-T DNA.

The first four steps constitute the synthesis part of theexperiment* These steps are designed to create colonies of JU_ coliwith a foreign gene expressed* Steps V and VI are designed to test thesynthesis steps and could be termed the analysis steps of theexperiment. The final steps continue the analysis and explore somerelated matters* The next section reviews each of these steps ingreater detail-.

111. 2 The Experimental Steps in Detail

I* Extract and purify the Phi -3-T DNA.

A. Verify purity (satisfactory level of protein contamination) bymeasuring ratio of UV absorption*

B. Verify a satisfactory degree of intact molecules and molecularweight using EM.

C Measure transforming ability in B^. subtilis. (Adequatetransforming activity is necessary to insure a reasonableexpectation of success in later steps.)

II Digest Phi-3-T DNA with the enzyme Ecoß and test fortransforming activity*

I

Experimental Reasoning 111.2 April 11, 1977

6

A* Perform complete digestion with Ecoß , (The motivation hereis to get Thy gene on a short segment since the complete phageis too long for introduction into the plasmid* The resultingsticky ends will be useful in the later ligation step*)

B* Use electrophoresis to obtain a cleavage pattern of thedigestion products and to estimate the molecular weight*

1* Compare the digestion products to published cleavagepattern*

2. Compare the electrophoresis and EM estimates for molecularweight* Discrepancy noted in molecular weight measurement*(EM measurement is higher than electrophoresis measurement

72 million vs.. 83 million*)

C* Generate and test hypotheses for molecular weight discrepancy.(See Chapter IV for a discussion of the reasoning behindthese hypotheses.) The following hypotheses were formed:

1 * Incomplete gel resolution*

2. Loss of small electrophoresis fragments migrating out ofgel.

3. Use of different standards for EM and gel-electrophoresis.For both techniques, molecular standards need to be usedwhich will be distinguishable from what is being measured.

4. Repetition in the phage genome. (This hypothesis wassuggested in the paper, but is hard to understand*)

D* Transforming activity of the Phi-3-T DNA was checked in B^_subtilis. It was observed that the transforming activity wasreduced by a factor of 1000 after complete Ecoß digestion.

E. Generate hypotheses to explain the loss of transformingactivity and test as follows:

1. Hypothesize that Ecoß damaged the Thy gene. Thishypothesis is disconfirmed by the fact that transformingactivity continues after complete digestion. Completedigestion was assumed by noticing no change in therestriction pattern after increasing the digestion time bya factor of 10, and increasing the enzyme concentration bya factor of 10.

2. Hypothesize edge effects. (This idea was discussed in athesis by Ron Harris but the effect has not yet beencompletely established*) This hypothesis was not pursued.

Experimental Reasoning 111*2 April 11, 1977

7

3* Hypothesize that transforming activity decreases if theDNA pieces are too small.

a* Decide to check the kinetics of Ecoß Phi-3-T DNAdigestion* Its transforming activity was plottedagainst digestion time* (A dependence was observed —thus supporting this hypothesis.)

b* Digest the fragments using BamHl - which cuts in fewerplaces* Again a ten-fold reduction in transformingactivity was observed. This further confirmed that theloss in transformational ability was due to the size ofthe pieces and not due to a property of EcoRT

F. Repeat the digestion step — limiting the process to partialdigestion.

Ill* Construct hybrid plasmids by mixing the partially EcoRT-digestedPhi-3-T DNA with Ecoßj-cut pSCIOI, followed by ligation with T4ligase*

IV* Transform !E_*_ coli with the plasmids* (See Chapter IV for thereasoning at this step.)

■pGrow colonies and select for Tc * A control is donesimultaneously with no plasmids added*

A

1 Hypothesize that most of these colonies correspond toplasmids which have simply reclosed under ligase*

2. Test for Phi-3-T DNA. Colony hybridizations indicates 8percent of the Tcr colonies have Phi-3-T DNA.

3. Select for Thy . Two colonies are Thy+*

B* Starting again, grow colonies and select for Thy+. Twocolonies were obtained* Again, a control is donesimultaneously with no plasmids added.

1* Select for Tcr . (Both colonies were Tcr )

V. Verify that the Thy+ gene is harbored on a plasmid* Otherplausible hypotheses should be tested* Possible hypotheses are:I)Thy is caused by reversion of Thy" to Thy+, or 2)the Phi-3-T thygene has integrated into the E_*_ coli chromosome*

A. Remove the hybrid plasmids from the E^ coli using a newlydeveloped curing technique* The cured E_*_ coli aredistinguished by being Tcs , All of the cured bacteria are also

April 11, 1977Experimental Reasoning 111*2

8

Thy"** This tends to rule out the alternate hypotheses since itconfirms that the Thy+ character is lost when the plasmid islost.

B. Propose further confirmation by reintroducing the plasmidsinto the cured bacteria * The same transformation frequency wasobserved in the cured bacteria as in the ones in which theplasmids had not been previously introduced.

C* Propose further confirmation step of introducing plasmid intoB* subtilis and testing for expression of Thy * (Done below*)

Vl* Use heteroduplex analysis to examine the molecular structure ofhybrid plasmids* (This time-consuming operation uses cRNA andEM.)

A* Look for segments common to all of the transformed plasmids.Observe segment A is in all transformed plasmids. Observeunexplained hairpin in pFT33 in segment corresponding tosegment A. This segment is longer and has distinct restrictionsites.

B. Hypothesize that segment A is from Phi -3-T. Confirm withheteroduplex mapping* Conclude that Thy is carried on segmentA.

C* Observe that segment A lies in two different orientations.Hypothesize that promoter control for Thy comes from Phi-3-TDNA and is part of segment A.

D. Propose confirming experiment with Col El-amp plasmid* Choiceof a different plasmid would disambiguate any special aspectsof the original vector which might be involved

VII. Transform B_;_ subtilis with the hybrid plasmids.-

A. Explore interesting theoretical question: Does the topology ofa plasmid (i.e. linear or circular) have any effect on thetransforming activity of DNA?

B* Test this question using BamHl to linearize the plasmid. Nodifference in transforming activity of hybrid plasmids noted.

VIII. Extend the colony hybridization technique to B*_ subtilis*Involves showing: (1) that DNA can be detected when it ispresent and (2) (specificity) that it will not be detected whenit is absent.

A* Show (1) by hybridizing cRNA from Phi-3-T with B* subtilisstrain lysogenized with the Phi-3-T,

Experimental Reasoning 111.2 April 11, 1977

9

B. Show (2) by demonstrating absence of hybridization with pSCIOICRNA*

IX* the (now tested) in situ hybridization procedure, performtne following measurements:

Test Phi -3-T pSCIOI

SB 168+

SB 168 (with lysogenic Phi-3-T)

58591 thy"++

58591 transformed with Phi-3-T58591 transformed with pFT23

58591 transformed with pFT24

E. coli (with pSCIOD

+

+

+

++Conclude that these results further confirm the hybridizationprocedure* Also recognizing that 58591 is a mutated Thy"derivative of SB 168 (a standard strain of B. subtilis with anintroduced lysogenic Phi-3-T), conclude that the mutagenesis hasdeleted sequences of Phi-3-T. 1 Also note that B. subtilistransformed with both pFT23 and pFT24 has kept the Phi^M" DNA butnot the pSCIOI DNA* (Suggests an intriguing selection process atthe molecular level*)

X* Reverse the Southern hybridization procedure — instead ofPM°^?PnSrSi^ thf plasmids and decking for hybridization withPfti-3-T DNA, digest and electrophorese the Phi-3-T and check forhybridization with the pFT's (using cRNA made from thepFT's.)Unexpectedly, hybridization was observed in several bands.(Expected only one band corresponding to Thy*) Postulate repetitionin the Phi-3-T genome. The experimenters attempted to confirm thisresult with heteroduplex analysis. This failed and the discrepancywas explained due to difficulties of this technique with largemolecules* c

übiished( ?ther hyPothe3es have been offere<* since the papers were

Results


10

Chapter IV

Review of Knowledge Used in This Experiment

The following paragraphs step through the experiment describedin Chapter 111 and emphasize the reasoning and the knowledge used tomake the decisions and interpretations* This knowledge is highlightedby indentations in the text below* No attempt has been made toclassify the knowledge here or to put it in a consistent form. Theeffort has been directed to writing down a first approximation to theknowledge with the intention of doing it more carefully at a later dateafter the scope of this knowledge is better understood* Thus, veryhigh level strategy knowledge has been freely intermixed with verycontext specific knowledge in a potpourri of facts, directives, andrules of inference. We begin with the selection of the experiment*

IV. I Proposing the Experiment

The knowledge used in proposing the basic experiment seems tobe difficult to encapsulate* Part of the problem is that theconsiderations can be very broad — involving political and regulatoryconsiderations as well as the directions of long term research goals*Another problem is that many of the considerations seem to be fairlyvolatile — what is a "hot" topic today will likely be less importanttomorrow. In what follows, we will present some of the knowledge whichseems to have been important in the proposal of this particularexperiment along with the caveat that the appropriate place forMOLGEN 's activity will undoubtedly be at a much lower level* No claimis made about the completeness of our characterization of the knowledgeused at this level.

Today, gene transfer is interestingespecially between species.

Today, it is interesting to study whether genecontrol signals from one species are operative inanother species.

These considerations derive from part of the long rangeresearch objectives of Lederberg's group* The particular experimentalso requires a choice of species and genes* As was stated earlier,

Experimental Reasoning IV* 1 April 11, 1977

11

attempts by this group to transfer a R*_ subtilis gene to E^ coli hadbeen unsuccessful. Two members of the group attended a conferl^c~e atCornell University in 1974 and brought back news of the bacteriophage-- Phi-3-T* This phage was especially interesting because of itsability to transform Thy B* subtilis to prototrophy* Additionally apublished reference was available which analyzed the EcoR T restrictionpattern of Phi-3-T . It was suggested that DNA from this 1 phage mightbe used as the source of the Thy gene in the gene transfer experiment.The following knowledge bears on that suggestion*

It is easier to clone genes which can beobtained in high concentration and purity.

Phage DNA can be obtained in highconcentration and purity.

A host species must also be chosen for a gene transferexperiment* Much experience has been gained working with E^ ooij andI*, subtilis. Because so much is known about the genetics andrequirements of these organisms, they are among the organisms of choicefor many genetics experiments* Some particularly relevant factsfollow*

When a species is available in strains withinactive genes , it may be a useful recipient foranalogous genes from other species in a genetransfer experiment*

There must be a means for incorporating theforeign DNA into the species so that it will bereproduced when the species grows.

Plasmids and lysogenic viruses are typicalvectors for introducing DNA between strains ofbacteria.

A large number of plasmids of jk coli havebeen characterized and are available.i

See [Wilson74]*

2 Phi-3-T DNA weighs 83 million, B* subtilis weighs 2,3 billion*It is reasonable to get phage DNA in concentrations of 10 pfu/ml(Plaque forming units per milliliter.) As indicated already, Phi-3-Twas known to have a transferable Thy region.

3 (that is, it is a well-characterized auxotroph)

IV. 1 April 11, 1977Experimental Reasoning

12

if

The reverse experiment, namely transferring Thy (or some othermarker) from jk coli to B*_ subtilis is also interesting* However, acloning vehicle (eg* vector) must be found* This is currently anactive area of interest*

Most recombinant DNA experiments (of which this is an example)involve the following technical choices:

I. Choice of a Method for cutting DNA* (Alternatives follow*)

A* One of the restriction enzymes*

B* Physical shearing*

II« Choice of a Method for joining DNA segments.

A* A Ligase (Usually either Jk coli ligase or T4-induced E. coliligase.)

B* The ( A: T) Terminal Transferase Method* This method involvesputting polyA and polyT sequences on the ends of the DNAsegments to be joined* Hydrogen bonds will form and hold thesegments together when they are mixed in solution*

C* Molecular Adapters* This is a recent technique suggested atthe last Miles conference. For situations where the DNA is cutby different restriction enzymes, the "sticky ends" of DNA willnot match properly and segments cannot usually be joined.Molecular adapters are short segments of DNA with alternatesticky ends corresponding to two different restriction enzymes*They may be used to splice together fragments resulting fromdigestion by different restriction enzymes*

lII* Choice of a vector for introducing the gene.

The following knowledge is relevant for these selections:

Restriction enzymes are useful for experimentswhere repeatable patterns need to be created* Theycan be used to create large numbers of identicalDNA segments*

Shearing can be used for situations where all

1V, 2 Establishing Some Subgoals


13

known restriction enzymes inactivate the desiredgene.

Shearing can be used to obtain a bank ofadjacent genes*

T4 Ligase is capable of joining segments whichare both flush-ended or segments with complementaryends ("sticky-ends")*

T4 Ligase of reliable purity is currentlyavailable in Lederberg's laboratory*

E*_ coli ligase (not T4) can be used to joinsticky ends, but not flush ended DNA.

A big advantage of the Terminal transferasemethod is that it insures that exactly one DNAregion will be inserted.

Molecular adapters were not available*

In the current experiment, the availability of T4 ligase andexperience with restriction enzymes led to the establishment of thefollowing subgoals for this experiment:

I* Isolate the gene on a segment of DNA of the right size havingsticky-ends left by a suitable restriction enzyme*

ll* Cut an appropriate plasmid with the same restriction enzyme so ithas the same sticky ends*

111 Ligate the gene segments and plasmid segments to create thehybrid plasmids*

IV. Transform the E*_ coli with the hybrid plasmids

In addition, the selection of a vector and restriction enzymehas to be made* The following knowledge appears to bear on theselection of the plasmid.

There should be good genetic markers for anygenetic transfer method*.

The vector should be small and easily

4 They now are available* See [Scheller77]

April 11, 1977Experimental Reasoning IV* 2

14

incorporated into a cell* (Many plasmids satisfythis requirement*)

It is a great advantage for purification stepsif there are a large number of copies of the vectorin each cell*

The vector should have a restriction site forone of the available restriction enzymes*

Punctuation signals must be available for theinserted gene* (These may be provided by theinserted segment or they can be near therestriction site of the vector*)

The utility of genetic markers on the vector derives from thefollowing considerations:

Three kinds of labeling are typically used inexperiments involving DNA manipulationDNA manipulationradioactive labeling, density labeling, andbiological labeling*

Biological labeling (genetic markers) offersthe advantage of amplification via growth andselection in a medium*

In this experiment the Tc gene on pSCIOI and pMB9 provide thismeans by way of the biological test for tetracycline resistance. Therestriction enzyme should be picked with the following considerations:

If a later ligation step is planned, therestriction enzyme should cut leaving complementary("sticky") ends*

If the restriction enzyme is known to cut(inactivate) a gene, it should probably not bechosen for use in an experiment to transfer thatgene *

A restriction enzyme should be chosen whichhas a recognition site compatible with the size ofsegments desired (eg* four, five, or six basepairs)*


15

1V,3 The First Part of the Experiment

As part of meeting the Subgoal I above, a sub-sub-goai is theextraction and purification of the Phi-3-T DNA* This should cause aselection among DNA-purification procedures. The method used in thisexperiment is a standard procedure. Some obvious but important rulesapply here:

Even if theory predicts something strongly,if a test is easy, do it* (You may be surprised*)'Verify any important step (e*g. purification)with further confirmational tests.

When there is a disagreement amongmeasurements, or between a local measurement and apublished measurement, find an alternate way to dothe measurement*

DNA purification steps should be checked forprotein contaminants, RNA, and degree of intactmolecules* Often it is appropriate to check thebiological (e*g* transforming) properties too.

An easy and fast test (OD 260/OD 280) isavailable for testing for the presence of proteinin DNA*

Electron microscopy can be used to test thedegree of intact molecules*

The habit of checking and verifying steps is übiquitous inexperimental procedures* In this experiment, the UV test was performedalthough it is so standard that it is usually not reported* Inaddition, the degree of intactness and molecular weight were measuredusing EM*

Following the digestion with Ecoß it is possible to observe arestriction pattern and to measure the molecular weight of the Phi-3-TDNA using gel electrophoresis. In this case, a discrepancy wasobserved between the EM measurement of the molecular weight and the gelelectrophoresis measurement. The following hypotheses were generatedto explain the discrepancy*

I. Incomplete gel resolution.

April 11, 1977Experimental Reasoning IV* 3

16

11. Loss of small DNA fragments which ran all the way through thegel.

111, Experimental error in measuring molecular weight.

IV. Repetition in the phage genome* (See Section IV*6*3«)

The following knowledge was relevant in generating thesehypotheses:

Incomplete gel resolution leads tounderestimation of molecular weight* (The usualassumption is that each band in the restrictionpattern contains the same amount of material andcorresponds to a single segment of the molecule* Ifresolution is incomplete, at least one bandactually corresponds to more than one segment andthe estimate of total molecular weight will be toolow* )

Loss of small fragments on the gel can lead toan underestimation of molecular weight* (Thismeans that the electrophoresis has been run so longthat the smaller fragments have migated all the waythrough the gel* )

Accuracy of gel electrophoresis for measuringmolecular weight in the linear part of its range isusually about two percent*

»

For both electrophoresis and EM, a DNAstandard must be run simultaneously to calibratethe measurement. In both cases it must be possibleto unambiguously distinguish the standard from themolecules being measured* Different standards weresuitable for these measurements* There has beensome disagreement about the value of the molecularweight for the standards*

No further experimental effort was spent on discriminatingbetween these hypotheses for the discrepancy* An alternate approach tomeasuring the molecular weight was proposed based on the followingknowledge :

5 SPPI DNA cut with Ecoß for electrophoresis, and pSCIOI(circular) for EM*

*Experimental Reasoning IV, 3 April 11, 1977

17

Incomplete gel resolution is often caused byhaving too many bands of DNA on the gel*

The number of bands in the gel is a functionof both the restriction enzyme used and the DNAitself* (It depends on the location and number ofrestriction sites.)

BamHl makes fewer cuts in Phi-3-T than Ecoß .(Four instead of about thirty* )

However, the fragments of Phi-3-T DNA left after digestion withBamHl are too large for accurate measurement with electrophoresis* Thusin this case, this idea does not provide a good alternate source forthe molecular weight measurement.

As noted earlier, it is often worthwhile to check thetransformational activity after a purification step and prior to acloning step* In this case, the completely Ecoß digested Phi-3-T DNAwas observed to transform B*_ subtilis, but at a 1000-fold decrease inefficiency compared to the uncut Phi-3-T DNA. The following hypotheseshave been suggested to explain the loss of efficiency:

I. Ecoßj cut the Thy gene — thus damaging its transforming ability..

ll* Transforming activity decreases if the DNA segments are toosmall*

lII* Edge effects

The following knowledge is relevant to the generation of thesehypotheses:

A gene which has been modified will usuallyfunction at an impaired efficiency.

Although the transformation process is notthoroughly understood, it is known that the processis influenced by DNA structural features.

A recent thesis by one of Lederberg's studentssuggested that some genes function less effectivelyif the DNA is cut near the gene *_See [Notani74] for a review of what is known about this

process*

7 This has not yet been demonstrated conclusively*

*

*


18

The first hypothesis was disconfirmed using the following knowledge*

If a gene can be shown to functional after itsDNA has been completely digested by a restrictionenzyme, the enzyme probably does not cut the gene*

If no change is observed in the restrictionpattern (in electrophoresis) for some DNA after aten-fold increase in digestion time and enzymeconcentration, the DNA ftmay be assumed to bedigested to completion*

The hypothesis about fragment size suggests a course of actionusing the knowledge that:

Sometimes experimental parameters can bechanged to maximize efficiency*

The completeness of digestion of DNA by anenzyme can be controlled by suboptimizing reactionrates. Many DNA segments after partial digestionof a restriction enzyme are larger than thesegments left after complete digestion*

Enzyme digestion conditions optimal forvarious purposes can be determined by studying theenzyme kinetics*

The experimenters decided to study the kinetics of transformingactivity versus digestion* Conditions appropriate for a 10-foldreduction in transforming activity were found and used previous to theligation step.

The hypothesis that the transforming activity depended on thesize of the segments (and not on some property of EcoR T ) was furtherconfirmed using the following knowledge:

Different restriction enzymes will cut DNAinto different numbers and sizes of pieces*_

Although this rule was cited in the papers, some exceptions toit are generally recognized* In the first place, there are inherentresolution limits which prevent some changes in restriction patternsfrom being observable in electrophoresis* Secondly, sometimesrestriction sites can be covered by trace proteins*


19

This leads to the confirmation step using BamHl as an alternaterestriction enzyme* The assumption behind this step is that Ecoß andBamHl will reduce the transforming activity by the same mechanism .

IV* 4 A Brief Recapitulation

To recapitulate some of the logic here, we started out with anexperiment for using Phi-3-T and E*_ coli based in part on some criteriafor interestingness of the the experiment and appropriateness for thespecies involved. Criteria such as availability led to the selection ofpSCIOI and the restriction enzyme Ecoß . The following subgoals wereestablished:

I* Isolate the Thy gene on a segment of DNA, that is, cut the Phi-3-TDNA so that the Thy gene will be located on a shorter segment ofDNA. (Use a restriction enzyme which leaves sticky ends.)

11. Cut the plasmid with the same restriction enzyme

111. Ligate a mixture of the cut Phi-3-T DNA and some cut pSCIOIplasmids resulting in some hybrid plasmids.

IV. Transform the jk coli with the hybrid plasmids

The first subgoal led us to a purification step and we pausedto test the purity of this step* An unexpected discrepancy inmolecular weight led us to some hypothesizing and checking* Then thenecessity to test the transforming activity of the Phi-3-T fragmentstook us afield because the activity was unacceptably low. This led tothe hypothesis that the loss of activity was related to the size of theDNA fragments. This hypothesis was tested and resulted in amodification to the plan — partially instead of completely digestingthe Phi-3-T DNA* (This modifies the second subgoal establishedpreviously*) We are now ready to pursue the last two subgoals.

1V. 5 The Second Half of the Experiment

g 7It is possible that the two restriction enzymes could bothreduce transforming activity of Phi-3-T DNA, but by different-mechanisms such as exonuclease contaminants* The considerations andhypotheses that were generated and tested on this topic were notreported in the papers*


20

The third subgoal is to ligate the plasmids* We have alreadycited some knowledge necessary for the selection of T4 ligase* Afterligation it is good practice to verify the success of the step* Thefollowing hypotheses could be tested:

I* The Ligase sealed the DNA*

ll* Phi-3-T DNA has been incorporated in the plasmids*

111. The Thy gene from Phi-3-T DNA has been incorporated in theplasmids*

The following knowledge is relevant to generating and testingthese hypotheses:

If the Ligase has sealed the DNA, it will formcovalently closed circles*

EM can easily and cheaply test whethercircular loops of DNA are in the sample*

10Ligation theory can be used to predict howmany of the plasmids will incorporate extra DNA*

Most of the plasmids will simply reclosewithout incorporating any Phi-3-T DNA.

If circular DNA in seen in EM, this willconstitute some evidence for successful ligation*

The following knowledge is relevant to the testing of the thirdhypothesis:

Heteroduplex mapping could be used to confirmthe incorporation of Phi-3-T DNA in the plasmids*

Heteroduplex analysis is most applicable to ahomogenous population of molecules*

See [Dugaiczyk7s]*

11 The ligation products are a mixture of linear and reclosedpSCIOI plasmids, Phi-3-T DNA fragments, and hybrid plasmids* Thebiological screening in the next step will help concentrate themolecules which include the Thy gene.

21

1V.5 April 11, 1977

If the plasmids are used to transform the E*coli and amplified by growth (the next step in thisexperiment), there will be a large number of hybridplasmids available for further testi ig»

These considerations led to using a simple EM test forsuccessful ligation in the current experiment* Testing of the secondand third hypotheses was deferred until after the transformation step.

The final subgoal in the basic experiment is to transform theE» coli with the hybrid plasmids. Like all important steps in anexperiment, transformation needs to be checked* The followinghypotheses about the results of the transformation could bedifferentiated:

I. No Ejt coli will have a Thy+ Tcr phenotype. (All colonies will beThy" Tc a , Thy" Tcr , or Thy Tc s ).

ll* Some Ex coll will have Thy* Tcr phenotype#

A* These E^ coli have no hybrid plasmids*

B* These JU. coli have hybrid plasmids*

1* The phenotype is not conferred by the plasmids.

a* Ej. coli showing a Thy* phenotype are revertants.

h* Ej. coli showing aTe phenotype are the products ofcontamination.

2» The Thy character is conferred by the plasmids.

Testing of the first hypothesis illustrates the importance ofbiological markers on the plasmid — Tcr in this case*

Biological markers have associated tests forfunction deficiencies* 1?Plus phenotypes can beselected for directly.

If bacteria can grow in a medium lacking in anessential nutrient, they are synthesizing it forthemselves*

If bacteria can grow in a medium with an

12 Minus phenotypes can be obtained using replica plating.

Experimental Reasoning

*

22

Experimental Reasoning IV. 5

antibiotic present (e.g. To), they are resistant toit at that concentration*

The Tc phenotypes could arise if none of the plasmids wereincorporated into the Ej. coli or if the gene for tetracyclineresistance has been inactivated* The Thy" phenotype would be expectedif either the Thymidine synthetase gene could not be activated in E*coli or if it had been lost or damaged during the preparations. The TcThy phenotype would have been unexpected, but might have resulted ifeither the tetracycline gene were damaged or the Thy gene had becomeincorporated by some unanticipated mechanism* As discussed in Section111*2, colonies with the Thy+ Tcr phenotype in fact were found*

The next hypothesis to be tested is whether the Thy Tccolonies contain hybrid plasmids* Three sources contribute evidencerelevant to this hypothesis — a biological argument about phenotypes,a colony hybridization step, and a heteroduplex analysis step.

The biological argument is somewhat indirect in that itprovides an opportunity for an easy counterexample* It is based on thefollowing knowledge:

Genes which are close together will tend tostay close together through transformation,transduction, conjugation, etc* Genes which arefar apart (eg* chromosomal vs plasmids) will stayfar apart^ and will generally not be co-transferred *

The fact that all of the E^ coli which are Thy* are also Tcr isindicative that the Thy phenotype is conferred by the plasmids*However, the appearance^ of any E± coli that were Thy+ Tc would havebeen quite unexpected and a serious challenge to the claim that thephenotype was conferred by hybrid plasmids* Genes linked in thismanner are especially useful in other experiments where there is nodirect way to select for the gene inserted on the plasmid*

The second source of evidence that there are hybrid plasmids is

There are exceptions to this rule that are not relevant here*For example, there is a sex-factor found in Hfr and F+ E. coli strains*These factors are conjugative (mobilizable) plasmids which can betransferred to F" recipients* In this experiment, all of the E, coliwere F"* )

They would most likely be due to a reversion of Thy" to Thy+*This was disconfirmed by the control plate as discussed below.

April 11, 1977

*

23

IV*5 April 11, 1977

from a colony hybridization step. This technique involves using RNApolymerase to make a template of radioactively labeled RNA which iscomplementary to the sequence we wish to detect* The bacteria aregrown on a filter, lysed (cell walls are burst), fixed, and then washedand hybridized with the cRNA. The cRNA will bind to any complementaryDNA on the filter and can be detected autoradiographicaliy. Again,this evidence does not conclusively demonstrate that there are hybridplasmids — only that DNA from Phi-3-T has been incorporated into theTc Thy colonies* The DNA could conceivably be incorporated in someother way.

The third source of evidence for hybrid plasmids is convincingbut time-consuming to perform — heteroduplex mapping. This stepdemonstrated the existence of hybrid plasmids and elucidated theirstructure* (If either of the previous tests had failed, it might nothave been worthwhile to perform this test.)

The next hypothesis to be tested is whether the Thy+ phenotypeis actually conferred by the plasmid. Some reason to test thisquestion is suggested by the following:

Genes can be carried on the bacterialchromosome or on extra-chromosomal DNA (episomes)such as plasmids or non-lysogenic phages*

Plasmids are occasionally picked up ascontaminants from the air or medium*

The hypothesis that the Thy+ character is the result of areversion is ruled out by a control plate where no plasmids have beenadded (and no Thy colonies appeared) along with the followingknowledge:

The Thy £&. coli in this experiment have beencharacterized as a deletion mutant*

Most deletion mutations very rarely revert.

Similarly, the hypothesis that the Tcr phenotype was the result ofcontamination was disconfirmed by an analogous control plate.

Disconfirming the specific hypotheses of other ways that thephenotypes could have been achieved does not prove that the phenotypeis necessarily the direct result of the hybrid plasmid because we havenot ruled out all possible contrary hypotheses* However, the followingidea does provide a method for a fairly direct demonstration.


24

1V.5 April 11, 1977

To show that A is the sole cause of B, Show(1) that when A then B and (2) that when not A thennot B.

The pSCIOI plasmids transform j^ coll withhigh efficiency*

The pSCIOI plasmids may be Hemoved from E.coli by a technique involving ethidium bromide,tetracycline, and ampicillin.

Techniques for removing or transformingbacteria with particular plasmids are expected tocontinue to work after small segments have beeninserted into the plasmids.

These considerations led to the experimental step of removingthe hybrid plasmids by the technique above. Eg. coli without theplasmid are identified by virtue of the fact that they are Tc3

* Thefact that the ceils which were Tc were also Thy" adds further credenceto the hypotheses above by establishing a linkage between the genes*Finally, the plasmids removed from the Thy+ Tcr cells showed a hightransforming efficiency for both the g± coli which never had the hybridplasmids, and for jSj. coli from which they had been removed* The lattercolonies were in every tested respect identical to the former* Inparticular, when the latter cells were mixed with the hybrid plasmids,they were transformed to Thy+ Tcr at the same high efficiency as theoriginal cells.

IV*6 Some Experimental Fishing Trips

At this point the experiment could very well have beenterminated* A great deal of evidence had confirmed the successfultransferring and expression of a gene from Phi-3-T to E^ coll* Thus,although some of the next steps in the research were partiallymotivated by a search for further confirming evidence that the Thygene was on the plasmid, the opportunity was taken to perform somesimple related experiments*

1V. 6.1 Transforming B* subtilis with the Hybrid Plasmids

Although the experiment as described above has already


25

Experimental Reasoning 1V.6 April 11, 1977

presented strong evidence for the successful transfer of a gene to E^coli, the following knowledge may have suggested some further steps.

A gene transferred to a new host can be testedfor modifications by reintroduction into its donor.

If a particular gene is known to function in aparticular species of bacteria, attempts to clonethat gene into a deficient strain will probablysucceed without complications involving geneexpression .

The Thy gene is carried by the bacteriophagePhi-3-T and expressed in its host B*. subtilis whenthe phage is incorporated lysogenically*

1* subtilis often incorporates DNA which itencounters in its environment (eg. it is highlytransformable)*

It would be interesting to know more aboutthe mechanism which B* subtilis uses to Incorporateand control foreign DNA*

thy- *Thi« f"««ests that it would be interesting to try to transformThy 1* subtilis with the hybrid plasmids (termed pFTs). Successfultransformation would also provide further confirmation that the Thy+

character of the transformed E*. ct&i was plasmid borne. The experimentwas^performed and it was observed the transformational activity was10 * At this point, the following knowledge appears to have beenactive*

(It is interesting to know what factorscontribute to the optimization of importantprocesses* )

The incorporation of DNA by bacteria is aprocess which is not well understood* It is notknown what structural features may influence thisprocess.

Ecoßj cut Phi-3-T DNA is linear*

15 I"There are sometimes complications involving dosage andlocations of promoter sites.

.

26

1V.6 April 11, 1977

The pFT's are circular and contain both pSCIOIDNA and Phi-3-T DNA*

The transforming activity of the Phi-3-T was-5

10 and the transforming activity of the circular-6

or BamHl cut pFT's was 10

If the pFT's are cut with Ecoßj the pSCIOI DNAwill become disconnected from tne Phi-3-T DNA(because the plasmids were ligated at EcoRT sites.)

This suggests that it would be interesting to decide whichfactors (differing between the pFT's and Ecoßj cut Phi-3-T DNA)determine the reduced transforming activity. The considerations aboveled to the experiment of cutting the pFT's with BamHl* No change intransforming activity was noted — although the cutting linearizes theplasmids* It may be hypothesized that the distinction between linearand circular DNA segments makes no difference to the B^. subtilis inthis case.

1V*6.2 Extending the Colony Hybridization Technique to B*subtilis

Rapid and reliable assays for importantproperties are worth developing.

In situ hybridization is a good assay forincorporated DNA and is an important assay forcloning experiments*

In situ hybridization has only been tested forDNA incorporated by E^ coli.

Sometimes techniques which are useful for onespecies can be extended without much effort toanother species*

These considerations suggest that the ijs aitu hybridizationtechnique could be extended to work for Ik subtilis as well as E*. coli.Some further experiments related to the step in the previous sectionwould then become easy to do*


*Experimental Reasoning 1V.6

27

To validate an assay, it is necessary todemonstrate both a capability for detection andspecificity ,When a phage is lysogenic with a bacterialhost, the phage DNA is incorporated into the thebacterial chromosome*

SBI6B is a standard and available strain of B.subtilis which can be lysogenized with Phi-3-T*

B^HThis k"owled8e suggests that the In situ hybridizationhvh^wce*

may *V? Sted if Phi ~3-T iy**«ni«ed SBT6T shows positivehybridization technique with Phi-3-T and negative hybridization with

When a new technique is validated, it isinteresting to try a variety of test cases* (Theymay suggest further research.)

58591 is a Thy" derivative of SB I6B availablein Lederberg's laboratory*

Th*

*m„

? faCt ' °f S6Ven 0f the test cases offered surprises.The first case was SBI6B tested (as was each test case) with Phi-3-T--nSCiniVRNAHand Pf" 1"^"1 om. As predicted, no homology withfvn

"

J f erV6d bUt the Phi"3"T ho*ology was a surprise (not yetexplained.) The second test case was the same as the validation testfor the method and offered no surprises* Next the mutated 58591 wastested* Interestingly, it showed no homologies with either cRNA whichsuggests that the mutation has deleted regions homologous to Phi-3-T.The next three test cases involved transformations of 58591 to Thy ~due to

PtL%hnd hi"3"T DNA * AS cxPected * homologies (presumablydue to the Thy gene) were detected in each case* Surprisingly no5K£ h«S f ?n/ith the °RNA fr°m PSCIOI ""

Pthat the b!ffth^Ji^!.* fl "**nta 0f DNA arisin« f"» the pSCIOI componentpfnln

VP ?! V <Further will be required to explicate this*)lirpri °aSe °f ** SSXL Wlth PSCI °I P lasmid s offered no

9nH . *° funanarize some of the knowledge implied by these test casesand conclusions:

f hThiS

DiS/f special case of the earlier rule* To show that A is

"h ?;wh°f \ ( : hat B measures A >» show (1) that when A then Band (2) that when not A then not B.

April 11, 1977

J

28

IV,6 April 11, 1977

When a bacterial strain is transformed by asegment of DNA and later tests show homologies foronly some parts of the introduced DNA, this isevidence for a selective process*

A mutation may involve structuralrearrangements, such as inversions, insertions,substitutions, or deletions of segments of DNA.

When a homology test, which is done for astrain and a mutated version of that strain, showsa loss of homology to some test DNA, thisconstitutes evidence for deletion of DNA by themutagenesis process.

A confirming test for a deletion mutant isabsence of reversion under selective pressure*

17The Southern method is useful for findingregions of homologies between two samples of DNA.

Homology tests between DNA samples A and B canbe performed by testing A against regions of B, orby testing B against regions of A.

In a previous step, the pFT's were tested against Phi-3-T DNAto provide evidence for insertion* Since it is just as easy to reversethe process, it was done and the surprising result was that pFT cRNA'shybridized with several Phi-3-T DNA bands. The following knowledge wasused in making some tentative conclusions:

When a segment of cRNA (or eDNA) showshomology to several different parts of a DNAmolecule, this constitutes evidence for repetitionin the molecule*

When a DNA molecule is completely digested bya restriction enzyme, different bands in therestriction pattern correspond to distinct regionsof the molecule.

17 See [Southern7s]*


1V.6.3 Back Hybridizing the pFT*s to Phi-3-T DNA

s

29

IV. 6 April 11, 1977

Heteroduplex mapping is useful for detectingmolecular rearrangements in DNA segments.

An alternative method to analyze repetitions — heteroduplexanalysis — revealed only one region of homology between the pFT andPhi-3-T. Furthermore, the hairpin region in pFT33 (which could arisefrom an inserted and inverted repeated sequence) remains unexplained.These discrepancies have yet to be explained*


<#


30

Chapter V

3&jie Thoughts About the Knowledge Used in This Experiment

Chapter IV stepped through the experiment and highlighted theknowledge that was active at critical decision points. In thischapter, a few general observations will be made about the knowledgeand its use in planning for this experiment.

V.l General Observations

There i& a large body of diverse knowledgeused in this experiment*

Approximately one hundred entries in the form of relevant factsor guidelines were highlighted in Chapter IV. When we actually try tofill -out this knowledge and formalize it as specific entries in aMOLGEN knowledge base, it will undoubtedly swell considerably. Thediversity of this knowledge suggests that a great deal of researcheffort in MOLGEN will be devoted to questions of representation.

Much £f the planning appears to be eventdriven*

Tne experiments described here reflect a combination of goaldriven behavior and event driven behavior. (The word "experiment"itself suggests that the procedure is somewhat tentative and intendedto elucidate an unknown effect or law.) If there were no goals,behavior might seem very erratic and follow no general course* Ifthere is no event driven component to the planning process, then theexperimental procedure must admit no feedback or change of plans as aresult of the observations. Thus, no advantage will be made offortunate observations. What is being suggested here is that theplanning in this experiment involved far more exploitation of eventsand changes of plan according to events than the authors hadanticipated. The importance of a combination of goal driven and eventdriven processes in problem solving has been discussed in theartificial intelligence literature *

See, for example, [Erman76] or [Engelmore77].

Experimental Reasoning V.I

31

April 11, 1977

One of the significant events in this experiment resulted fro-the observation of the low transforming efficiency ot the £oR cut MAD^rff^; "S l6d t0 30ne hyP°êSe3 about the optima1 size ofnLuIT. f°f transformation, some conflrmational measurements? andoartiallv „,S f"erat,io" of the high level plan. (The DNA was onlySEh ?£'„? B^*\ 15?taad °f belng o°mPieteiy digested before ligationwith the plasmids.) Finally, some events did not lead to changes in theexperiment but did contribute to the wealth of conclusions which couJdTlftlrl?t\Jr.Z T°PLc ' the *-«"«"« that inserted segments"coZlultr, y»£d plasmi?s

"«-~ "*«t*d in opposite directions led to aoSCIOI T,m t r°m°ter 00ntr01 ' Sil»ila'-ly. the observation thatpSCIOI DNA was not incorporated into 8,. aiMUiS. led to a tentativeresearch Unlike* the'^0" "d a" a~a *>' "«rtiSorbi:rv:uon U"af„coe

ottan

eti:ip3acetr

ceV

da :ion """Bt -Nation, this

One explanation for some of the event-driven character of theZZIT"V3 , faCt th3t Plannlng must take place even though theknowledge is incomplete. For example, when it was proposed that anBamHl TnTT^ f »01*^ «*** could be obtained by usingBamHJ instead of Ecoß to digest the Phi-3-T DNA beforfelectrophoresis, the number^f restriction sites and size of the Ph^!IrfESTt" St" "0t k"OWn in advan°c'e ' Thus planni"8 must beOf the kno^LT^ and ,°!!eCked aft6r co"Pletion. The incompleteness?sk„ *

knowledge may take the form of unknown properties of thelaboratory techniques as well as unknown attributes of DNA structuresFor example in a recent laboratory meeting, the question [as askedwhether a particular enzyme used in an experiment was precessive ??Vwhether it tends to remain attached to a single molecule^! S In f^t'explalnL^'and^lhr606831:6 ' ♦* faUUre in the coSld beIttereT experimental procedure could be appropriately

UIS. geneticist Is opportunistic and tries toMake discoveries.

, MThuf' not only is the Planning process largely event driven butsometimes steps are taken somewhat outside the plan of the experiment

reflet afh

PoSSibly, interesting observation. ?his kind of behaviorChlte ?h. °onve"ie?oe °f "kin* certain interesting observationswhile the equipment is set up. Often this is done to verify theobsâllonsê:"^ °f a" eTriaental Step > but -mlurmesy S:possTblmles s««» to correspond more to fishing for interestingpossibilities. One example was the linearization of the hybridplasmids with BamHl to see if topology was important in theincorporation of the plasmids in I* safetiiia. Another^xa^ple was thetransformation of B, safety by the hybrid plasmids. (BeUer evidence

V *-

Experimental Reasoning V.l April 11, 1977

32

was already available that the Thy gene was plasmid borne and that thetransformation of E* coli had been achieved.) Several examples of thissearch for possibly interesting observations were presented in SectionIV. 6*

Hypothesis formation is an important activitin planning experiments.

Hypothesis formation is especially critical when anexperimental prediction fails -- that is — when the unexpected isobserved in an experiment. For example, a difference in homology led toan hypothesis about the mutagenesis of 58591* In this case, hypothesisformation could be described by a single rule of evidence* Other casesof hypothesis formation are more involved* In such cases, theknowledge of the limitations and effects of laboratory techniques islikely to play a role in the formation of hypotheses* One example ofthis has already been cited — the hypothesis that the lowtransformational activity of completely EcoRT digested Phi-3-T DNA wascaused by the small size of the remaining DNA segments. Similarly,when the discrepancy in molecular weight was discovered early in theexperiment, hypotheses had to be generated which could explain thedifferences among different sources for the measurement*

In many of these situations, the generation of the hypothesescan be understood as a systematic checking of the assumptions used inthe model to make the disconfirmed prediction. Systematic generationof plausible hypotheses is clearly one of the most important processesin the experimental science. Some effort in this direction could proveinteresting for MOLGEN*

It is often difficult to determine when thereis enough evidence*

From one point of view, this observation is equivalent to theprevious one. There is sufficient evidence when all competinghypotheses have been ruled out* Practically, there is no way to becertain that ail plausible hypotheses have been considered, and thereare many hypotheses which are too farfetched to merit seriousexamination* The problem arises from the fact that often no laboratorytechnique is available to directly measure the item of interest.Measurements have to be interpreted and techniques are subject tooccasional failure. This generally leads to a fairly conservativeapproach to experimental proof so that several confirming measurementsare made when the result is important*

V

Experimental Reasoning V*l April 11, 1977

33

Perhaps the best example of rather careful confirmation anddifferentiation between hypotheses is the reasoning involved inverifying the success of transforming E^. coli* It was deemedinsufficient to show merely that the E*. coli were Thy+ . Rather theevidence was sifted to see precisely whether it indicated (1) that someE* coli were Thy and (2) that the plasmid contained an insert of Phi--3-T DNA and (3) that the Thy character was conferred by the plasmid.This careful testing reflects an ability to differentiate betweenhypotheses which could explain some of the tests individually withoutnecessarily confirming the central hypothesis.

Thus, a great deal of evidence was gathered in support of themost important conclusion — the successful transfer of the Thy gene toIx coli. Comparatively less effort was spent in checking the validityof the colony hybridization technique for B*. subtilis. even thoughseveral surprises were found when it was used investigatively.Interestingly, a failure in the established heteroduplex analysistechnique was postulated when it failed to confirm a prediction aboutrepetition in the phage genome* This suggests (1) that evenestablished techniques are occasionally suspect and (2) that the mostimportant conclusions merit the most redundant checking.

V.2 Some Proposed Important Parameters

In reviewing the knowledge expressed in Chapter IV, it appearsthat some concepts have broad application.

Importance Determines how worthwhile it is topursue certain objectives at this time(like AM's "interestingness").

Effort How expensive or time-consuming it isto pursue certain objectives. (Resourcelimitations*)

Certainty Determines believability of inferencesand observations (like MYCIN *scertainty factors).

2 See [Lenat76]*

3 See [Shortiiffe76]*

■"*

<m

V.2 April 11, 1977Experimental Reasoning

34

Is the experiment safe and does it fitwithin regulatory guidelines?

Safety

Availability Availability of materials is a majorconsideration*

This list is by no means complete and several specializationsmay prove useful. For example, one can distinguish between items whichare important in a particular context, and those which are importantgenerally. Something can be important in the sense of beingfundamental or in the sense of being novel or suprising* Planningdecisions may make considerations of more domain dependent notions —for example yield, purity, or shelf life.

Many of these parameters appear in a variety of decision makingcontexts* For example, importance and effort are useful in deciding onvery long range objectives as well as short term and event drivendecisions. Important conclusions merit the most careful (redundant)verification. Unpredicted observations must be examined for theirimportance (interestingness, novelty, and impact on genetic theory)before effort is spent to explain them or perhaps even base furtherresearch on them*

Many combinations of these parameters are especially relevantin decision making. For example, it is probable that resources shouldbe allocated to an experiment which has a high importance and a loweffort* An example of this is the linearization of the hybrid plasmidsin this experiment. If the verification of a hypothesis has a higheffort while a low effort counterexample can be tested, thecounterexample may be tested first even if it is unlikely* This is thesource of many of the "controls" which are routinely run* In caseswhere both the importance and the estimated effort are high, resourcesare less likely to be allocated " An example of this would be a studyinto the intriguing mechanism by which B^ subtilis used some form ofmolecular selection to reject the pSCIOI part of the hybrid plasmid butaccept the Phi-3-T part of the plasmid. (This study could be agenetics thesis itself.) When hypotheses are being verified, animportant hypothesis will merit considerable verification even if itsplausibility is already quite high. A less I important hypothesis ofequal certainty will get somewhat weaker confirmation* An unimportanthypothesis of high certainty will probably get little attention* Otherinteresting combinations of parameters include such things as importantand easy but unsafe experiments* Much needs to be learned about whenit is optimal for considerations of these different parameters to enterinto the planning process* It seems that the characterization of theseparameters for decision making could be an interesting and significantpart of MOLGEN research*_

A cost/benefit analysis is required in general*.

■* "■- V


35

As previously mentioned, this paper made no attempt tocategorize the rules used in planning the experiment. Rather, therationale behind each experimental step was described by invoking rulesat a variety of levels of detail. For example, in Section 1V* 6.2, arule specifying the importance of verifying an important result* isimmediately followed by another rule which details the presence of aparticular bacterial strain in the lab stock collection. This jumpingfrom extremely high-level rules of scientific decision making to veryspecific statements of available resources needs to be examined in muchgreater detail.

It appears to be of critical importance to systematize therules used in planning experiments. We have summarized a subset ofthis domain in describing the cloning experiment* However, many otherpossible strategies exist and many other experimental goals arepossible. A proposed method of gaining knowledge about alternatives isby examining other experiments in detail. For this, a coherentstructure needs to exist in order to delineate what rules, and at whatlevels of detail, are needed* To this end, a more formal notation ishighly desirable. Our next step would seem to be generating aclassification scheme for the rules used in this experiment with a viewtowards expanding the structure to including rules from other publishedexperiments.

V.M Concluding Remarks

A surprise in working on this set of experiments was the extentto which considerations outside the usual set of hierarchical planningideas entered into the experimental planning. Far more of the decisionmaking was event driven than had been anticipated and it remains to beseen whether this situation is characteristic of the current moleculargenetics domain. In addition, some basic scientific activities — suchas hypothesis formation and testing, which had not previously receivedmuch attention in MOLGEN — now appear to be quite important*

One of the important benefits of this exercise has been to makeexplicit some of the domain knowledge* One clear lesson from thisexercise has been a greater realization of the importance of includingsome knowledge about biological function* Finally, more work needs tobe done to categorize and represent the knowledge described in thisreport and to determine compact subsets of genetic knowledge which willprovide the richest material for artificial intelligence research*

v*3 Rule Classifications

'«*>


36

Appendix I

Bibliography

[Dugaiczyk7s] Dugaiczyk A., Boyer H.W., Goodman H.M., Ligation of Ecoß.Endonuclease-generated DNA Fragments into Linear and CircularStructures, Journal of Molecular Biology, pp 171-184 (1975)

[Ehrlich76] S.D* Ehrlich, H* Bursztyn-Pettegrew, I. Stroynowski, and J.Lederberg, Expression of the thymidylate synthetase gene of theBacillus sijbtljis bacteriophage Phi-3-T in Escherichia coll.Proceedings of the National Academy of Sciences USA. 73:11, pp,4145-4149, (November 1976)

[Ehriich77] S.D* Ehrlich, H. Bursztyn-Pettegrew, I* Stroynowski, and J*Lederberg, Cloning of the Thymidylate Synthetase Gene of the PhagePhi-3-T, Proceedings of the X Miles Conference, in press.

[Engelmore 77] Engelmore R.S*, Nii H.P., A Knowledge-Base System for theInterpretation of Protein X-Ray Crystallographic Data, ComputerScience Department Report No. Stan-CS-77-589, Stanford University

[Erman76] Erman L.D., Overview of the HEARSAY Speech UnderstandingResearch, Working Papers in Speech Recognition -IV- the HEARSAY IISystem, Carnegie-Mellon University, Computer Science Speech Group,(1976)

[Lenat76] Lenat D.B*, AM: An Artificial Intelligence Approach toDiscovery in Mathematics as Heuristic Search, PhD Thesis ComputerScience Department, Stanford University (1976)

[Martm77] Martin N*, Friedland P., King J., Stefik M.J., KnowledgeBase Management for Experiment Planning in Molecular Genetics,Heuristic Programming Project HPP-77-19, Computer ScienceDepartment, Stanford University (1977) (Submitted to the FifthInternational Conference on Artificial Intelligence.)

[Notani74] Notani N.K*, Setlow J.K., Mechanism of BacterialTransformation and Transfections, in Cohn W.E. (cd.) Progress inNucleic Acid Research in Molecular Biology, J4 (1974)

[Platt64] Piatt J.R*, Strong Inference, Science 146, pp 347 (1964)

[Scheiler77] Scheller R.H., et. al., Chemical Synthesis of RestrictionEnzyme Recognition Sites Useful for Cloning, Science 196:177 (1977)

ma.

Experimental Reasoning V.il

37

April 11, 1977

[Shortliffe76] Shortliffe E mvptu.

Consultations, New York: American ElsevierP£er"baBed

Biology I^PP 503-517 (1975)Eleotrophoresi« " Journal of Molecular

lSte^ll^TX;iV Lrtain

GeNnet

A *?~ °f *«— *~ "-blemReport

8 77-596T Computer Science" designing System, CS(1977)

computer Science Department, Stanford University

[Wilson7l] Wilson, G.A., Williams M t bCharacterization of Tempest' to'tlJV^' H *W" a"d F 'E ' y°un«,by the Restriction Endonuclease Ecoß ?TS °f„ Sa£iiAlia SutUlUBacteriophages, J. Virol UMOl* ( W*J Afferent Temperate

bibliography - stanford universityqs400mf2512/qs400mf2512.pdfuse of different standardsfor em and...

Documents