## User: Steven Lakin

Steven Lakin •

**1.4k**- Reputation:
**1,390**- Status:
- Trusted
- Location:
- Fort Collins, CO, USA
- Last seen:
- 5 months, 2 weeks ago
- Joined:
- 3 years, 11 months ago
- Email:
- S***********@colostate.edu

Veterinary/PhD student at Colorado State University. My research applies statistics and machine learning to problems in biomedical science and public health. My computational experience is in data mining/data science, statistics, machine learning, bioinformatics algorithms, programming, genomics, and transcriptomics. Wetlab experience includes NGS library prep, PCR, and culture-based methods. Interests include pathogenesis of infectious disease, antimicrobial resistance, public health, and epidemiology.

https://github.com/lakinsm

#### Posts by Steven Lakin

<prev
• 117 results •
page 1 of 12 •
next >

1

vote

1

answer

956

views

1

answers

... DESeq2 as well as other differential gene expression pipelines call some kind of glm as a subroutine (lm would be inappropriate here, since it's only linear. You want something more robust, based on count distributions); DESeq is largely based on the negative binomial distribution and will take int ...

written 2.3 years ago by
Steven Lakin •

**1.4k**0

votes

1

answer

956

views

1

answers

... In your metadata file, record the percentage of cancer cells for each sample and have that as one of your factors for analysis. Regression is used in differential gene expression analysis to calculate eventual log-fold change, so if you regress on your % of cancer cells, you'll hopefully capture th ...

written 2.3 years ago by
Steven Lakin •

**1.4k**0

votes

2

answers

2.8k

views

2

answers

... To elaborate on this; don't do this.
Do instead '0 + mainEffect1 + covariate1) ...

written 2.3 years ago by
Steven Lakin •

**1.4k**1

vote

2

answers

2.8k

views

2

answers

... Standard regression analysis for count data is based on modeling expression values as a function of explanatory variables that are part of the experimental design:
y ~ 0 + effect1 + effect2 + (1 | effect 3)
For example. Typically, we want to explain the expression values as a function of some sys ...

written 2.3 years ago by
Steven Lakin •

**1.4k**0

votes

2

answers

2.8k

views

2

answers

... Right, which is what we product based on standard logic. Let me know if you're interested in the standard experimental designs. ...

written 2.3 years ago by
Steven Lakin •

**1.4k**0

votes

2

answers

1.2k

views

2

answers

... What you did in your very first design (not this answer), where you found no DE genes, was correct. What you've done above is incorrect; stick with your intuition on this one. It's not uncommon for groups to be insignificant on DE analysis, especially with cancer expression analysis. ...

written 2.3 years ago by
Steven Lakin •

**1.4k**3

votes

2

answers

1.2k

views

2

answers

... It's nice that you were able to see that the pairwise analyses in DESeq and limma are equivalent to regression and then fitting on the contrasts by hand; they are exactly the same process, only DESeq and limma handle them for you in the background if you specify the model in the way you originally d ...

written 2.3 years ago by
Steven Lakin •

**1.4k**0

votes

2

answers

2.8k

views

2

answers

... I think I understand what you're trying to do now. Apologies; it is the first time I've seen someone request this in bioinformatics, as it is somewhat strange to model this way. What you will have to do is construct your model matrix like so (this assumes your probe matrix is, by native format, in ...

written 2.3 years ago by
Steven Lakin •

**1.4k**1

vote

2

answers

2.8k

views

2

answers

... Programs like Limma force the gene expression values to be the response variable because that is the correct way to model it:
lmFit(probe_matrix, design = model.matrix( ~ 0 + Disease phenotype + covariate )
This will calculate regression coefficients for disease phenotype and the covariate as ...

written 2.3 years ago by
Steven Lakin •

**1.4k**0

votes

2

answers

1.0k

views

2

answers

... Hi Brian. Do you happen to have a repository for the node pruning tool? It sounds useful. ...

written 2.3 years ago by
Steven Lakin •

**1.4k**#### Latest awards to Steven Lakin

Popular Question
13 months ago,
created a question with more than 1,000 views.
For Does nhmmer look at reverse complements?

Teacher
13 months ago,
created an answer with at least 3 up-votes.
For A: Retrieving specific columns from clinicaltrial.gov using R

Appreciated
2.1 years ago,
created a post with more than 5 votes.
For C: Fisher's exact test gives p-value 0

Good Answer
2.1 years ago,
created an answer that was upvoted at least 5 times.
For A: Is it OK to select a small subset of expression data for statistical testing?

Teacher
2.3 years ago,
created an answer with at least 3 up-votes.
For A: Retrieving specific columns from clinicaltrial.gov using R

Teacher
2.3 years ago,
created an answer with at least 3 up-votes.
For A: Retrieving specific columns from clinicaltrial.gov using R

Scholar
2.3 years ago,
created an answer that has been accepted.
For A: Allele count from factor variable in R

Teacher
2.3 years ago,
created an answer with at least 3 up-votes.
For A: Retrieving specific columns from clinicaltrial.gov using R

Appreciated
2.3 years ago,
created a post with more than 5 votes.
For C: Fisher's exact test gives p-value 0

Scholar
2.3 years ago,
created an answer that has been accepted.
For A: Allele count from factor variable in R

Centurion
2.3 years ago,
created 100 posts.

Scholar
2.3 years ago,
created an answer that has been accepted.
For A: Allele count from factor variable in R

Good Answer
2.3 years ago,
created an answer that was upvoted at least 5 times.
For A: Is it OK to select a small subset of expression data for statistical testing?

Appreciated
2.3 years ago,
created a post with more than 5 votes.
For C: Fisher's exact test gives p-value 0

Guru
2.3 years ago,
received more than 100 upvotes.

Teacher
2.4 years ago,
created an answer with at least 3 up-votes.
For A: Retrieving specific columns from clinicaltrial.gov using R

Commentator
2.4 years ago,
created a comment with at least 3 up-votes.
For C: Fisher's exact test gives p-value 0

Appreciated
2.6 years ago,
created a post with more than 5 votes.
For C: Fisher's exact test gives p-value 0

Scholar
2.8 years ago,
created an answer that has been accepted.
For A: Allele count from factor variable in R

Appreciated
2.9 years ago,
created a post with more than 5 votes.
For C: Fisher's exact test gives p-value 0

Supporter
2.9 years ago,
voted at least 25 times.

Use of this site constitutes acceptance of our User
Agreement
and Privacy
Policy.

Powered by Biostar
version 2.3.0

Traffic: 880 users visited in the last hour