Course in Molecular Neuroanatomy Group Project

I was lucky enough to visit Okinawa again for the Course in Molecular Neuroanatomy. Thanks to OIST and Allen Brain Atlas.

As a tutor I had to facilitate a group project. I was tasked with the auditory cortex and after a quick google search I found this paper:
Gene Expression Identifies Distinct Ascending Glutamatergic Pathways to Frequency-Organized Auditory Cortex in the Rat Brain, by Storace, Higgins, Chikar, Oliver, and Read

It was all done in rat it seems. I sent out the paper to the group and we took a look at some of their findings in the paper in the Allen Mouse datasets. We were happy to find agreement in the Allen mouse gene expression and the connectivity atlases. Limited to the results regarding VGLUT1 and cortex <-> thalamus connections.

Here's the powerpoint presentation that summarizes our work.

Updates (based on response from Douglas Storace):

So it's pretty easy to do a quick spatial correlative search in the Allen Brain atlas, to find other genes that have a similar pattern.

So the one gene we mention with a similar pattern is Kcnma1. I just tried to reproduce that on the Allen website and it's not so easy.

Here's the steps:

1. search for slc17a7
2. checkbox the saggital assay (the first one)
3. goto the correlation box on the right and leave basic cell groups checked, and also checkbox thalamus
4. click search in the correlation box
5. Kcnma1 shows up as 6th.

It seems that search wasn't spatially restricted to thalamus as we have in the slides. You could try variations on that to find more genes - like selecting all three Slc17a7 assays.

When you get those lists, you should go through them by hand and take a look for gradients. We did this quickly for our project so the confidence we have in Kcnma1 is limited.

Also, one neat thing we found in the human data is that VGLUT2 is highly expressed in subcortical auditory regions like cochlear nuclei, Inferior Colliculus and others.. it's expressed all over.. but it's
relatively higher in those regions.

Axon guidance and SNPs

This is warning to those doing pathway type analysis on SNPs extracted from Illumina 660 type chips. Specifically when looking at brain phenotypes.

Recently, I took a quick and fast approach of grabbing a bunch of genes from some top (but non significant) SNP's from an analysis of fMRI data. I didn't expect any GO groups to be significant, so I was shocked to see axon guidance on top with a super significant p-value (10^-17 after correction). There's a bunch of other GO groups too that come out on top too (see below). We were doing brain stuff so we liked the GO groups but were suspicious for a number of reasons.

Here's a quote that sums up the problem:
"Such an analysis of gene set enrichment is based on the assumptions that all genes are sampled independently from each other with the same probability. These assumptions are violated with data from GWA studies as (i) longer genes usually have more SNPs resulting in a higher probability of being sampled and (ii) overlapping genes are sampled in clusters (Holmans et al., 2009)."

That text is from:
Gowinda: unbiased analysis of gene set enrichment for genome-wide association studies
Robert Kofler and Christian Schlötterer

One quick test I did was to run another gene ontology analysis where the genes are sorted according do how many SNP's on the genotyping chip. That confirms the problem. Here's some of the top groups from a quick GOrilla analysis

GO Group	FDR q-value
cell adhesion	3.43E-017
biological adhesion	2.04E-017
neuron projection guidance	1.39E-017
axon guidance	1.04E-017
single-organism process	1.84E-015
synaptic transmission	3.16E-014
single-organism cellular process	1.20E-013
cellular component movement	1.40E-013
single-multicellular organism process	2.30E-011
ion transport	2.51E-011

This shows a pretty clear bias and I'm guessing it's in part biological. I'd reckon that lots of variation in neuronal guidance genes allows a diversity of brain wiring.

Next step was to hookup Gowinda for the human data and redo the analysis. The result was no significant GO groups.

Several files are needed Gowinda, I put them online if anyone wants to do a similar analysis with human data.

We noticed this and fixed it, but I wonder how many other papers might have been tricked by this. One that comes to mind is this paper:
A genomic pathway approach to a complex disease: axon guidance and Parkinson disease
Timothy G Lesnick, Spiridon Papapetropoulos, Deborah C Mash, Jarlath Ffrench-Mullen, Lina Shehadeh, Mariza de Andrade, John R Henley,Walter A Rocca, J. Eric Ahlskog, Demetrius M Maraganore

with this follow-up (among others):
Neither Replication nor Simulation Supports a Role for the Axon Guidance Pathway in the Genetics of Parkinson's Disease
Yonghong Li, Charles Rowland, Georgia Xiromerisiou, Robert J. Lagier, Steven J. Schrodi, Efthimios Dradiotis, David Ross, Nam Bui, Joseph Catanese, Konstantinos Aggelakis, Andrew Grupe, Georgios Hadjigeorgiou

I can't say exactly what might explain the differences between those two as they did a more complex analysis than just looking for pathway enrichment for certain SNPs. The first paper does have very significant p-values (10^-51) which can be a red flag. It could be that a small enrichment for brain genes, for the phenotype could be exaggerated by the huge bias in the chips. I didn't spend much time on this though, I just wanted to post it online incase anyone else runs into it.

Update: I noticed a similar problem when running GO analysis on Illumina 450k methylation data. In that case the number of CpG sites per gene is not even across GO groups. I ended up doing an empirical type analysis which seemed to work - the super low p-values disappeared.

Tools for molecular neuroanatomy

This is just a quick listing of tools I often recommend to students at the Course in Molecular Neuroanatomy. They are mainly for dealing with many genes.

From slides:

• Open office
• Libre office
• Google drive
• CSV transpose tool for Expression.csv

Finding out when and where genes are expressed:

• Excross
• mouse genes as input
• returns when and where a list of genes is expressed
• find out which genes are in other Allen datasets
• wait for images to load
• reload page for new input
• HBAset
• specific expression for many genes in the 6 Allen HBA donors
• returns where a list of genes are specifically expressed
• under construction
• treeview button doesn't work right in chrome, try firefox

I made the above two so if you find them helpful or are using them for analysis then contact me for extensions and help.

Lapis

• extracting gene symbols from text
• visual xml parsing, method:
• click on the edit simultaneously button
• select gene symbol or text of interest
• copy and paste into a new file inside lapis
• copy again and paste in excel or somewhere else

VennMaster

• comparing gene lists
• making venn/Euler diagrams
• input is a text file with tab separated gene then group entries
• use a spreadsheet tool to make list, then paste into text file
• use control+c to copy genes in mac osx

Hypergeometic test or Fisher's exact test

• used to test if the overlap between two gene sets is significant

Gene ontology analysis

• GOrilla for quick analysis
• needs a list of background genes
• human
• mouse
• ErmineJ for more control of settings

Gene network

• Cytoscape
• ReactomeF1 app

Ortholog conversion, dbOrtho

• easiest to just set input+output to gene symbol

Extracting gene lists from PDF files - pdftohtml maybe useful. 

Thanks to Christos Gkogkas for suggesting I keep track of these.

Low SES and the brain

I'm interested in psychoneuroimmunology and just recently I made a website for running sets of genes against the neuroanatomically impressive Allen Human Brain atlas (also works on the Allen fetal brain data).

This lets me find out which brain regions show specific expression for a given set of genes when compared to the rest of the genome and brain.

I dropped in the genes from "Low early-life social class leaves a biological residue manifested by decreased glucocorticoid and increased proinflammatory signaling" by Greg Miller and colleagues. I combined the over and under expressed lists to see where they are turned on across 416 brain regions (nonparametric, AUC/ROC analysis, 1-6 donors per region). The p-values have been FDR corrected for number of regions.

The area I'd guess at is the hypothalamus and 7 of it's 29 regions are significant with all 7 up regulated. Going deeper it's the paraventricular nucleus and supraoptic nucleus in the anterior region. In the tuberal region the arcuate and lateral subregions have higher expression.  The paraventricular nucleus (ranked #1 of 416) and supraoptic nucleus are good to see because they link to oxytocin and vaspressin. Both of those have been linked to maternal related behaviour - and so has early life SES.

Here's the top 12 of the 416 regions:

Rank	Name	Donors	Q-value  (FDR corrected p-value)
1	Supraoptic Nucleus, Left	5	0.000004
2	Supraoptic Nucleus, Right	2	0.000004
3	central grey of the pons, Right	1	0.00001
4	Substantia Nigra, pars reticulata, Left	6	0.0001
5	central grey of the pons, Left	2	0.0003
6	arcuate nucleus of medulla, Right	2	0.001
7	Globose Nucleus, Right	2	0.001
8	Paraventricular Nucleus of the Hypothalamus, Left	5	0.001
9	Emboliform Nucleus, Right	2	0.001
10	globus pallidus, internal segment, Right	2	0.001
11	pontine raphe nucleus, Left	3	0.001
12	Paraventricular Nucleus of the Hypothalamus, Right	1	0.002

It seems the cerebellar cortex turns down these genes while the cerebellar nuclei (globus, emboliform nucleus) turns them on. I'm not sure how to interpret that, but it might be cell type driven.

Periaqueductal gray/central grey of the pons is also strong. That's nice to see because it contains vasopressin and oxytocin receptors and it's in the CAN/central autonomic network. A few of these overlap with the CAN network but I haven't put a p-value on that.

Again, I just wanted to put this neat little result out there online as it's a distraction that has some value I reckon. If anyone is interested I can use different gene lists, narrow to major brain divisions and try it on fetal data (12-16 post conception weeks).

Update: It seems vasopressin and oxytocin are linked to the immune response (from rat studies). Vasopressin and oxytocin neurons activate when LPS is given, and it seems strong in the supraoptic nucleus.

Update: Dr. Miller noted that I'm using gene expression results from blood so it's pretty indirect (I should've mentioned this). My response is that it's rough, but I'm trying to get a perspective into the brain. There's psychiatric studies where they use blood gene expression for looking at Parkinson's and schizophrenia cases. Also, over 90% of genes are turned on the brain.

Update: I found out the hypothalamus is unique in it's connection with the blood. It lacks a normal blood brain barrier it seems. I will have to look at unrelated studies of blood and see if any list of blood expressed genes will express specifically in the hypothalamus.

SAT1 (spermidine/spermine N1-acetyltransferase 1), suicide and skin color

The other day a frieind of mine - Nancy Yu posted a link on google+ to a recent article "A Blood Test for Suicide?"

After reading the news version, it seems like a pretty solid finding came out in a few studies - SAT1. In terms of molecular function the article states "which is involved in cellular damage and stress". I'm guessing the referenced articles have more to say about Sat1 but I decided to crunch some data instead of reading.

By data, I mean prenatal RNA-seq gene expression data from the Brainspan consortium. I used only the pre-natal samples which include 233 samples from 18 donor brains. For crunching, I have a co-expression method setup to look for co-expressing GO groups and all I have to do is give it a gene and run it. It's guilt by association.

So out of curiosity I do this every now and then. This potential suicide biomarker seemed like a good target.

I can explain more if someone requests. I just wanted to put this out there as it's a small finding. If someone is interested I will give all the details and run on the adult samples or other Allen data.

I ran SAT1 and new GO group appeared on top: "developmental pigmentation" (39 genes), it's defined as "The developmental process that results in the deposition of coloring matter in an organism, tissue or cell.". From my quick look, this appears to be a novel finding.

For this method, I'm still not sure how to put a p-value on things but in both methods the SAT1 to developmental pigmentation co-expression signal is significant after correction for the 4,032 GO groups tested. What is most astounding, is that SAT1 has a higher co-expression score with the developmental pigmentation genes than any other gene tested - including the 39 genes that are already annotated as developmental pigmentation. That's as specific as it gets.

Also, "lamellipodium membrane" comes out for cellular component.

Next, I decided to look up relationship between suicide and pigmentation as I have some faith in this result. I was happy to find "Suicide rate and skin color" by Voracek M. It has a short abstract:

"In two previously described samples of 53 and 43 nations, higher suicide rates of men, women, and the elderly population consistently corresponded to lighter skin color. This ecological (aggregate-level) association was independent of national differences in affluence, as measured by per capita Gross Domestic Product."

I stopped at that point as I don't have time to dive into the literature and I couldn't find that as a PDF. Instead I decided to just post this and email it out to Dr. Gustavo Turecki, Dr. Martin Voracek and Dr. Alexander Niculescu III. Hopefully they can take it further.

The four top co-expressing genes are:
RAB27A(0.48), EDNRB(0.47), MYO5A(-0.47), HPS6(-0.46)

The top ten list is:

1. developmental pigmentation
2. regulation of cation channel activity
3. fatty acid transmembrane transport
4. regulation of transmembrane transport
5. bone remodeling
6. pigment cell differentiation
7. calcium-mediated signaling
8. bone resorption
9. regulation of viral genome replication
10. amine transport

Update: Dr. Niculescu responded via email and mentioned that "Melanoma Signalling" shows up in the ingenuity pathway analyses they did. That's in Table S4 in their paper - Discovery and validation of blood biomarkers for suicidality. That's using a bunch of genes for input - not just SAT1, so that's nice to hear about as I'm not sure if I would have found it by reading through the paper.

Also, here's a good quote from the Le-Niculescu et al. paper "SAT1-overexpressing mice had alterations in their polyamine pool, hair loss, infertility and weight loss." (with links to two references).

Update 2: I managed to get the Voracek full text paper. It's very short and provides country level correlations for mostly european nations. Again, I'm not sure what to make of it.  Here's a quote: "This pattern suggests that the correspondence of higher suicide rates with lighter skin color and vice versa across nations is unlikely to be due to national differences in affluence" (Voracek, 2006). It would be nice to see plots of the Voracek data. It cites another paper that links eye color to method of suicide: "Tattoos, eye color and method for suicide." I can't find the PDF for that so I should stop the distraction as I doubt there is more individual level data.

Update 3: I recently heard that light therapy works for PTSD and there's one pilot study in depression. I emailed the first and last authors of the pilot study. I got a quick response from Dr. Hamblin who noted that sunlight is beneficial but the light therapy penetrates much deeper. Still, this a lot of speculation but I wonder if SAT1 is absorbing sunlight that's needed to keep the brain normal.

Update 4: I just read a writeup (needs reg) on this talk at the APA 2014 conference: Abstract NR2-186. "The Correlation of Excessive Indoor Tanning With Depression and Suicidal Behavior Among Adolescents: Results From the 2011 Youth Risk Behavior Survey" by Kelly E. Taylor, also Dr. Gathright is mentioned. Looks like there is a huge association for depression but also several suicide related measures (odds ratio for treatment for suicide attempt is 13.1!, N=15,425).

Update 5: This not strongly related but I wanted to point out a study that linked disruption of circadian rhythm with depression. In this study there was data from 20 suicide cases. It's a bit indirect, but light is strongly related to circadian rhythm.

Update 6: On Dr. Turecki's suggestion, I took a look at the postnatal BrainSpan data and 'developmental pigmentation' doesn't show up. I'm surprised to see the change, one thing I noticed is that SAT1 is negatively correlated with the majority of genes in the postnatal data. The top ten from the postnatal data (345 samples) are:

1. cerebellar cortex formation
2. regulation of receptor-mediated endocytosis
3. cerebellar Purkinje cell layer morphogenesis
4. membrane depolarization
5. cerebellar Purkinje cell layer development
6. cellular copper ion homeostasis
7. membrane depolarization involved in regulation of action potential
8. membrane depolarization involved in regulation of cardiac muscle cell action potential
9. cellular response to prostaglandin stimulus
10. positive regulation of protein ubiquitination involved in ubiquitin-dependent protein catabolic process

One note is that the ethnicity distribution between prenatal and postnatal samples is very different, so I should take a closer look at that when I have time.


Update 7: Neat paper looking at UV exposure - seems it's addictive. "Skin β-Endorphin Mediates Addiction to UV Light"