Gene Expression analysis

Get data

Υπάρχουν αρκετές βάσεις δεδομένων που μπορούμε να κατεβάσουμε δεδομένα έκφρασης γονιδίων. Σε αυτή τη σειρά μαθημάτων θα χρησιμοποιήσουμε το “Gene Expression Omnibus” ή αλλιώς GEO. Η διεύθυνση του αναγράφεται παρακάτω.

https://www.ncbi.nlm.nih.gov/geo/

Από το GEO μας ενδιαφέρουν τα datasets, και ειδικότερα αυτά που πέρα από τα raw δεδομένα (CEL files), δίνουν και δεδομένα τα οποία έχουν προ-επεξεργαστεί και ειναι έτοιμα για ανάλυση. Αυτά τα βρίσκουμε στο λινκ “Datasets” που υπάρχει σε κάποιο σημείο στην αρχική σελίδα του GEO.

Στην συνέχεια έχουμε την δυνατότητα να χρησιμοποιήσουμε

τον κωδικό του dataset αν τον γνωρίζουμε
κάποια keywords.

Εμείς θα χρησιμοποιήσουμε το keyword “Smoking”, ώστε να δούμε datasets τα οποια σχετίζονται με τις επιδράσεις του καπνίσματος.

Ετσι, θα κατεβάσουμε το dataset “GDS3709 Cigarette smoke effect on the oral mucosa”. Για αρχή ψάχνουμε ένα dataset με σχετικά απλό πειραματικό σχεδιασμό ώστε να μπορέσουμε να μελετήσουμε τα στατιστικά τμήματα της αναλυσης.

Μπορούμε να κατεβάσουμε το dataset χρησιμοποιώντας το λινκ στο “DataSet SOFT file”.

Ειδάλλως, κάντε copy το λινκ και χρησιμοποιήστε μία από τις δύο γραμμές στην R

## get data
# 1ος τρόπος
download.file(url = "ftp://ftp.ncbi.nlm.nih.gov/geo/datasets/GDS3nnn/GDS3709/soft/GDS3709.soft.gz", destfile = "./GDS3709.soft.gz")
# 2ος τρόπος
system("wget ftp://ftp.ncbi.nlm.nih.gov/geo/datasets/GDS3nnn/GDS3709/soft/GDS3709.soft.gz")

Ο πρώτος τρόπος, ειναι κώδικας R. έχει όμως το μειονέκτημα ότι πρέπει να δοθεί το destfile. Ο 2ος τρόπς ειναι στην ουσία system-call της εντολής wget.

Διαχείρηση των δεδομένων

Αφού κατέβει, το πρώτο βήμα ειναι να κάνουμε decompress στο αρχείο.

gunzip GDS3709.soft.gz

Αν στην συνέχεια το ανοίξουμε θα δούμε ότι οι αρχικές γραμμές ειναι ως εξής:

^DATABASE = Geo
!Database_name = Gene Expression Omnibus (GEO)
!Database_institute = NCBI NLM NIH
!Database_web_link = http://www.ncbi.nlm.nih.gov/geo
!Database_email = geo@ncbi.nlm.nih.gov
!Database_ref = Nucleic Acids Res. 2005 Jan 1;33 Database Issue:D562-6
^DATASET = GDS3709
!dataset_title = Cigarette smoke effect on the oral mucosa
!dataset_description = Analysis of oral mucosae from 40 cigarette smokers and 40 age and gender matched never-smokers. Results provide insight into the carcinogenic effects of cigarette smoke.
!dataset_type = Expression profiling by array
!dataset_pubmed_id = 20179299
!dataset_platform = GPL570
!dataset_platform_organism = Homo sapiens
!dataset_platform_technology_type = in situ oligonucleotide
!dataset_feature_count = 54675
!dataset_sample_organism = Homo sapiens
!dataset_sample_type = RNA
!dataset_channel_count = 1
!dataset_sample_count = 79
!dataset_value_type = count
!dataset_reference_series = GSE17913
!dataset_order = none
!dataset_update_date = May 13 2010
^SUBSET = GDS3709_1
!subset_dataset_id = GDS3709
!subset_description = female
!subset_sample_id = GSM447400,GSM447401,GSM447402,GSM447403,GSM447405,GSM447411,GSM447413,GSM447415,GSM447416,GSM447418,GSM447422,GSM447424,GSM447425,GSM447427,GSM447428,GSM447429,GSM447430,GSM447431,GSM447432,GSM447434,GSM447435,GSM447440,GSM447442,GSM447444,GSM447448,GSM447449,GSM447450,GSM447451,GSM447452,GSM447458,GSM447461,GSM447462,GSM447463,GSM447464,GSM447467,GSM447468,GSM447469,GSM447472,GSM447473
!subset_type = gender
^SUBSET = GDS3709_2
!subset_dataset_id = GDS3709
!subset_description = male
!subset_sample_id = GSM447398,GSM447399,GSM447404,GSM447406,GSM447407,GSM447408,GSM447409,GSM447410,GSM447412,GSM447414,GSM447417,GSM447419,GSM447420,GSM447421,GSM447423,GSM447426,GSM447433,GSM447436,GSM447437,GSM447438,GSM447439,GSM447441,GSM447443,GSM447445,GSM447446,GSM447447,GSM447453,GSM447454,GSM447455,GSM447456,GSM447457,GSM447459,GSM447460,GSM447465,GSM447466,GSM447470,GSM447471,GSM447474,GSM447475,GSM447476
!subset_type = gender
^SUBSET = GDS3709_3
!subset_dataset_id = GDS3709
!subset_description = cigarette smoke
!subset_sample_id = GSM447401,GSM447404,GSM447406,GSM447407,GSM447409,GSM447411,GSM447412,GSM447413,GSM447415,GSM447416,GSM447425,GSM447426,GSM447430,GSM447433,GSM447435,GSM447439,GSM447440,GSM447441,GSM447443,GSM447444,GSM447445,GSM447446,GSM447448,GSM447449,GSM447450,GSM447452,GSM447453,GSM447455,GSM447456,GSM447458,GSM447459,GSM447461,GSM447464,GSM447466,GSM447468,GSM447470,GSM447472,GSM447474,GSM447475
!subset_type = agent
^SUBSET = GDS3709_4
!subset_dataset_id = GDS3709
!subset_description = control
!subset_sample_id = GSM447398,GSM447399,GSM447400,GSM447402,GSM447403,GSM447405,GSM447408,GSM447410,GSM447414,GSM447417,GSM447418,GSM447419,GSM447420,GSM447421,GSM447422,GSM447423,GSM447424,GSM447427,GSM447428,GSM447429,GSM447431,GSM447432,GSM447434,GSM447436,GSM447437,GSM447438,GSM447442,GSM447447,GSM447451,GSM447454,GSM447457,GSM447460,GSM447462,GSM447463,GSM447465,GSM447467,GSM447469,GSM447471,GSM447473,GSM447476
!subset_type = agent
^DATASET = GDS3709
#ID_REF = Platform reference identifier
#IDENTIFIER = identifier
#GSM447401 = Value for GSM447401: Smoker female study #107; src: Smoker female buccal mucosa
#GSM447411 = Value for GSM447411: Smoker female study #20; src: Smoker female buccal mucosa
#GSM447413 = Value for GSM447413: Smoker female study #205; src: Smoker female buccal mucosa
#GSM447415 = Value for GSM447415: Smoker female study #210; src: Smoker female buccal mucosa
#GSM447416 = Value for GSM447416: Smoker female study #216; src: Smoker female buccal mucosa
#GSM447425 = Value for GSM447425: Smoker female study #26; src: Smoker female buccal mucosa
#GSM447430 = Value for GSM447430: Smoker female study #2; src: Smoker female buccal mucosa
#GSM447435 = Value for GSM447435: Smoker female study #33; src: Smoker female buccal mucosa
#GSM447440 = Value for GSM447440: Smoker female study #41; src: Smoker female buccal mucosa
#GSM447444 = Value for GSM447444: Smoker female study #48; src: Smoker female buccal mucosa
#GSM447448 = Value for GSM447448: Smoker female study #51; src: Smoker female buccal mucosa
#GSM447449 = Value for GSM447449: Smoker female study #53; src: Smoker female buccal mucosa
#GSM447450 = Value for GSM447450: Smoker female study #55; src: Smoker female buccal mucosa
#GSM447452 = Value for GSM447452: Smoker female study #5; src: Smoker female buccal mucosa
#GSM447458 = Value for GSM447458: Smoker female study #68; src: Smoker female buccal mucosa
#GSM447461 = Value for GSM447461: Smoker female study #71; src: Smoker female buccal mucosa
#GSM447464 = Value for GSM447464: Smoker female study #77; src: Smoker female buccal mucosa
#GSM447468 = Value for GSM447468: Smoker female study #88; src: Smoker female buccal mucosa
#GSM447472 = Value for GSM447472: Smoker female study #93; src: Smoker female buccal mucosa
#GSM447400 = Value for GSM447400: Never smoker female study #105; src: Never smoker female buccal mucosa
#GSM447402 = Value for GSM447402: Never smoker female study #10; src: Never smoker female buccal mucosa
#GSM447403 = Value for GSM447403: Never smoker female study #110; src: Never smoker female buccal mucosa
#GSM447405 = Value for GSM447405: Never smoker female study #11; src: Never smoker female buccal mucosa
#GSM447418 = Value for GSM447418: Never smoker female study #248; src: Never smoker female buccal mucosa
#GSM447422 = Value for GSM447422: Never smoker female study #265; src: Never smoker female buccal mucosa
#GSM447424 = Value for GSM447424: Never smoker female study #269; src: Never smoker female buccal mucosa
#GSM447427 = Value for GSM447427: Never smoker female study #272; src: Never smoker female buccal mucosa
#GSM447428 = Value for GSM447428: Never smoker female study #274; src: Never smoker female buccal mucosa
#GSM447429 = Value for GSM447429: Never smoker female study #299; src: Never smoker female buccal mucosa
#GSM447431 = Value for GSM447431: Never smoker female study #3; src: Never smoker female buccal mucosa
#GSM447432 = Value for GSM447432: Never smoker female study #30; src: Never smoker female buccal mucosa
#GSM447434 = Value for GSM447434: Never smoker female study #32; src: Never smoker female buccal mucosa
#GSM447442 = Value for GSM447442: Never smoker female study #43; src: Never smoker female buccal mucosa
#GSM447451 = Value for GSM447451: Never smoker female study #57; src: Never smoker female buccal mucosa
#GSM447462 = Value for GSM447462: Never smoker female study #72; src: Never smoker female buccal mucosa
#GSM447463 = Value for GSM447463: Never smoker female study #74; src: Never smoker female buccal mucosa
#GSM447467 = Value for GSM447467: Never smoker female study #84; src: Never smoker female buccal mucosa
#GSM447469 = Value for GSM447469: Never smoker female study #8; src: Never smoker female buccal mucosa
#GSM447473 = Value for GSM447473: Never smoker female study #95; src: Never smoker female buccal mucosa
#GSM447404 = Value for GSM447404: Smoker male study #113; src: Smoker male buccal mucosa
#GSM447406 = Value for GSM447406: Smoker male study #12; src: Smoker male buccal mucosa
#GSM447407 = Value for GSM447407: Smoker male study #14; src: Smoker male buccal mucosa
#GSM447409 = Value for GSM447409: Smoker male study #19; src: Smoker male buccal mucosa
#GSM447412 = Value for GSM447412: Smoker male study #203; src: Smoker male buccal mucosa
#GSM447426 = Value for GSM447426: Smoker male study #271; src: Smoker male buccal mucosa
#GSM447433 = Value for GSM447433: Smoker male study #31; src: Smoker male buccal mucosa
#GSM447439 = Value for GSM447439: Smoker male study #40; src: Smoker male buccal mucosa
#GSM447441 = Value for GSM447441: Smoker male study #42; src: Smoker male buccal mucosa
#GSM447443 = Value for GSM447443: Smoker male study #44; src: Smoker male buccal mucosa
#GSM447445 = Value for GSM447445: Smoker male study #49; src: Smoker male buccal mucosa
#GSM447446 = Value for GSM447446: Smoker male study #4; src: Smoker male buccal mucosa
#GSM447453 = Value for GSM447453: Smoker male study #61; src: Smoker male buccal mucosa
#GSM447455 = Value for GSM447455: Smoker male study #64; src: Smoker male buccal mucosa
#GSM447456 = Value for GSM447456: Smoker male study #65; src: Smoker male buccal mucosa
#GSM447459 = Value for GSM447459: Smoker male study #6; src: Smoker male buccal mucosa
#GSM447466 = Value for GSM447466: Smoker male study #81; src: Smoker male buccal mucosa
#GSM447470 = Value for GSM447470: Smoker male study #9; src: Smoker male buccal mucosa
#GSM447474 = Value for GSM447474: Smoker male study #96; src: Smoker male buccal mucosa
#GSM447475 = Value for GSM447475: Smoker male study #98; src: Smoker male buccal mucosa
#GSM447398 = Value for GSM447398: Never smoker male study #100; src: Never smoker male buccal mucosa
#GSM447399 = Value for GSM447399: Never smoker male study #102; src: Never smoker male buccal mucosa
#GSM447408 = Value for GSM447408: Never smoker male study #17; src: Never smoker male buccal mucosa
#GSM447410 = Value for GSM447410: Never smoker male study #1; src: Never smoker male buccal mucosa
#GSM447414 = Value for GSM447414: Never smoker male study #21; src: Never smoker male buccal mucosa
#GSM447417 = Value for GSM447417: Never smoker male study #22; src: Never smoker male buccal mucosa
#GSM447419 = Value for GSM447419: Never smoker male study #251; src: Never smoker male buccal mucosa
#GSM447420 = Value for GSM447420: Never smoker male study #255; src: Never smoker male buccal mucosa
#GSM447421 = Value for GSM447421: Never smoker male study #260; src: Never smoker male buccal mucosa
#GSM447423 = Value for GSM447423: Never smoker male study #267; src: Never smoker male buccal mucosa
#GSM447436 = Value for GSM447436: Never smoker male study #35; src: Never smoker male buccal mucosa
#GSM447437 = Value for GSM447437: Never smoker male study #36; src: Never smoker male buccal mucosa
#GSM447438 = Value for GSM447438: Never smoker male study #38; src: Never smoker male buccal mucosa
#GSM447447 = Value for GSM447447: Never smoker male study #50; src: Never smoker male buccal mucosa
#GSM447454 = Value for GSM447454: Never smoker male study #63; src: Never smoker male buccal mucosa
#GSM447457 = Value for GSM447457: Never smoker male study #66; src: Never smoker male buccal mucosa
#GSM447460 = Value for GSM447460: Never smoker male study #70; src: Never smoker male buccal mucosa
#GSM447465 = Value for GSM447465: Never smoker male study #80; src: Never smoker male buccal mucosa
#GSM447471 = Value for GSM447471: Never smoker male study #91; src: Never smoker male buccal mucosa
#GSM447476 = Value for GSM447476: Never smoker male study #9; src: Never smoker male buccal mucosa
!dataset_table_begin

Επίσης στο τέλος-τέλος (τελευταία γραμμή) εχει την παρακάτω γραμμή.

!dataset_table_end

Πέρα από αυτές τις γραμμές, οι υπόλοιπες γραμμές έχουν την μορφή πίνακα. Δηλαδή

ID_REF  IDENTIFIER      GSM447401       GSM447411       GSM447413       GSM447415       GSM447416       GSM447425       GSM447430       GSM447435       GSM447
1007_s_at       MIR4640 1124    1196    982.8   1075    1114    1302    1279    1210    1199    1172    1245    1307    1263    1282    1191    1189    1202
1053_at RFC2    203.3   181.5   229.6   160.4   209.5   191.9   180.6   204.3   173.5   249.8   213.3   191.8   202.7   187     191.2   173.4   162.7   172.1
117_at  HSPA6   53.4    55.6    49.04   39.1    51.94   69.8    52.44   51.25   47.07   109.6   48.1    43.84   49.29   59.64   39      50.07   49.22   43.33

Μπορούμε να πάρουμε το τμήμα του αρχείου που θέλουμε είτε

χειροκίνητα, ανοίγοντας το σε έναν editor και σβήνοντας τις γραμμές που αρχίζουν με !, #, ή ^
με κάποια unix εντολή πχ

grep -e '^[\^\#\!]' -v GDS3709.soft > GDS3709.soft.clean

Εδώ απλά χρησιμοποιούμε την grep ώστε να αγνοήσει τις γραμμές που ξεκινούν με τους παραπάνω χαρακτήρες. Οι υπόλοιπες γραμμές αποθηκεύουνται στο GDS3709.soft.clean.

Ανάλυση και πληροφορίες για το dataset.

Στην GEO datasets σελίδα, υπάρχει ένα “κουμπί”, που γράφει “Data subsets” και μας πληροφορεί για το τι είδους data υπάρχουν σε κάθε στήλη. Έτσι…

Samples Factors Title
gender  agent 
GSM447401   female  cigarette smoke Smoker female study #107
GSM447411   female  cigarette smoke Smoker female study #20
GSM447413   female  cigarette smoke Smoker female study #205
GSM447415   female  cigarette smoke Smoker female study #210
GSM447416   female  cigarette smoke Smoker female study #216
GSM447425   female  cigarette smoke Smoker female study #26
GSM447430   female  cigarette smoke Smoker female study #2
GSM447435   female  cigarette smoke Smoker female study #33
GSM447440   female  cigarette smoke Smoker female study #41
GSM447444   female  cigarette smoke Smoker female study #48
GSM447448   female  cigarette smoke Smoker female study #51
GSM447449   female  cigarette smoke Smoker female study #53
GSM447450   female  cigarette smoke Smoker female study #55
GSM447452   female  cigarette smoke Smoker female study #5
GSM447458   female  cigarette smoke Smoker female study #68
GSM447461   female  cigarette smoke Smoker female study #71
GSM447464   female  cigarette smoke Smoker female study #77
GSM447468   female  cigarette smoke Smoker female study #88
GSM447472   female  cigarette smoke Smoker female study #93
GSM447400   female  control Never smoker female study #105
GSM447402   female  control Never smoker female study #10
GSM447403   female  control Never smoker female study #110
GSM447405   female  control Never smoker female study #11
GSM447418   female  control Never smoker female study #248
GSM447422   female  control Never smoker female study #265
GSM447424   female  control Never smoker female study #269
GSM447427   female  control Never smoker female study #272
GSM447428   female  control Never smoker female study #274
GSM447429   female  control Never smoker female study #299
GSM447431   female  control Never smoker female study #3
GSM447432   female  control Never smoker female study #30
GSM447434   female  control Never smoker female study #32
GSM447442   female  control Never smoker female study #43
GSM447451   female  control Never smoker female study #57
GSM447462   female  control Never smoker female study #72
GSM447463   female  control Never smoker female study #74
GSM447467   female  control Never smoker female study #84
GSM447469   female  control Never smoker female study #8
GSM447473   female  control Never smoker female study #95
...

Έτσι, ξέρουμε ότι έχουμε male, female δειγματα και καπνιστές και μη καπνιστές. Το φύλο και το αν καπνίζει κάποιος ειναι ανεξάρτητες μεταβλητές που επηρεάζουν την εξάρτημένη μεταβλητή, η οποία ειναι το επίπεδο έκφρασης του κάθε γονιδίου. Άρα εμείς θα προσπαθήσουμε να “εξηγήσουμε” την τιμή έκφρασης με βάση του φύλο του ατόμου και το αν καπνίζει κάποιος. Αρχικά θα τα δούμε αυτά χωριστά και στην συνέχεια θα τα συνδυάσουμε.

Όμως, αρχικά, ας δούμε μερικά στοιχεία για την κατανομή των τιμών έκφρασης στο κάθε άτομο

raw.data = read.table("GDS3709.soft.clean", header = TRUE)
dim(raw.data)

## [1] 54675    81

data <- raw.data[,-c(1,2)]
dim(data)

## [1] 54675    79

## let's get only the females. The first 39 samples are females

female.data <- data[,1:39]
label <- rep(c("smoker", "control"), c(19, 20))
label

##  [1] "smoker"  "smoker"  "smoker"  "smoker"  "smoker"  "smoker"  "smoker" 
##  [8] "smoker"  "smoker"  "smoker"  "smoker"  "smoker"  "smoker"  "smoker" 
## [15] "smoker"  "smoker"  "smoker"  "smoker"  "smoker"  "control" "control"
## [22] "control" "control" "control" "control" "control" "control" "control"
## [29] "control" "control" "control" "control" "control" "control" "control"
## [36] "control" "control" "control" "control"

## this is very skewed towards small values
boxplot((female.data))

## we need to obtain the log of this
boxplot(log(female.data))

Gene Expression analysis - microarrays

Pavlos Pavlidis

3/22/2021

Get data

Διαχείρηση των δεδομένων

Ανάλυση και πληροφορίες για το dataset.